Merge pull request #4423 from joycech333/feat/inception-llm-service

feat: add Inception LLM service with Mercury 2 support

.agents/skills/changelog

@@ -0,0 +1 @@
+../../.claude/skills/changelog

.agents/skills/cleanup

@@ -0,0 +1 @@
+../../.claude/skills/cleanup

.agents/skills/code-review

@@ -0,0 +1 @@
+../../.claude/skills/code-review

.agents/skills/docstring

@@ -0,0 +1 @@
+../../.claude/skills/docstring

.agents/skills/pr-description

@@ -0,0 +1 @@
+../../.claude/skills/pr-description

.agents/skills/pr-submit

@@ -0,0 +1 @@
+../../.claude/skills/pr-submit

.agents/skills/update-docs

@@ -0,0 +1 @@
+../../.claude/skills/update-docs

.claude-plugin/marketplace.json

@@ -0,0 +1,27 @@
+{
+  "name": "pipecat-dev-skills",
+  "owner": {
+    "name": "Pipecat"
+  },
+  "metadata": {
+    "description": "Development workflow skills for contributing to the Pipecat project",
+    "version": "1.0.0"
+  },
+  "plugins": [
+    {
+      "name": "pipecat-dev",
+      "description": "Development workflow skills for contributing to the Pipecat project",
+      "version": "1.0.0",
+      "source": "./",
+      "skills": [
+        "./.claude/skills/changelog",
+        "./.claude/skills/cleanup",
+        "./.claude/skills/code-review",
+        "./.claude/skills/docstring",
+        "./.claude/skills/pr-description",
+        "./.claude/skills/pr-submit",
+        "./.claude/skills/update-docs"
+      ]
+    }
+  ]
+}

.claude/settings.json

@@ -0,0 +1,5 @@
+{
+  "attribution": {
+    "commit": ""
+  }
+}

.claude/skills/changelog/SKILL.md

@@ -0,0 +1,61 @@
+---
+name: changelog
+description: Create changelog files for important commits in a PR
+---
+
+Create changelog files for the important commits in this PR. The PR number is provided as an argument.
+
+## Instructions
+
+1. Skip changelog for: documentation-only, internal refactoring, test-only, CI changes.
+
+2. First, check what commits are on the current branch compared to main:
+   ```
+   git log main..HEAD --oneline
+   ```
+
+3. For each significant change, create a changelog file in the `changelog/` folder using the format:
+   Allowed types: `added`, `changed`, `deprecated`, `removed`, `fixed`, `security`, `performance`, `other`
+   - `{PR_NUMBER}.added.md` - for new features
+   - `{PR_NUMBER}.added.2.md`, `{PR_NUMBER}.added.3.md` - for additional entries of the same type
+   - `{PR_NUMBER}.changed.md` - for changes to existing functionality
+   - `{PR_NUMBER}.fixed.md` - for bug fixes
+   - `{PR_NUMBER}.deprecated.md` - for deprecations
+   - `{PR_NUMBER}.removed.md` - for removed features
+   - `{PR_NUMBER}.security.md` - for security fixes
+   - `{PR_NUMBER}.performance.md` - for performance improvements
+   - `{PR_NUMBER}.other.md` - for other changes
+
+4. Each changelog file should at least contain a main single line starting with `- ` followed by a clear description of the change. No line wrapping.
+
+5. If the change is complicated, changelog files can have indented lines after the main line with additional details or code samples.
+
+6. Use ⚠️ emoji prefix for breaking changes.
+
+7. **Write changes in user-facing terms first.** Lead with what users of the framework will notice: new APIs, changed behavior, new parameters, fixed bugs they might have hit, etc. Implementation details (internal refactoring, how something is wired up under the hood) can be included as secondary context after the user-facing description, but should never be the *only* content of a changelog entry when there is a user-visible effect.
+
+   **Good** (user-facing first, implementation detail as context):
+   ```
+   - Turn completion instructions now persist correctly across full context updates when using `system_instruction`. Previously they were injected as a context system message, which caused warning spam and didn't survive context updates.
+   ```
+
+   **Bad** (implementation detail only, no user-facing framing):
+   ```
+   - Fixed turn completion instructions being injected as a context system message instead of using `system_instruction`.
+   ```
+
+   Ask yourself: "If I'm a developer building on Pipecat, what would I notice changed?" Start there.
+
+## Example
+
+For PR #3519 with a new feature and a bug fix:
+
+`changelog/3519.added.md`:
+```
+- Added `SomeNewFeature` for doing something useful.
+```
+
+`changelog/3519.fixed.md`:
+```
+- Fixed an issue where something was not working correctly in some user-visible scenario. The root cause was an internal implementation detail.
+```

.claude/skills/cleanup/SKILL.md

@@ -0,0 +1,312 @@
+---
+name: cleanup
+description: Review, refactor, document, and validate code changes in the current branch
+---
+
+# Code Cleanup Skill
+
+The **Code Cleanup Skill** reviews, refactors, and documents code changes in your current branch, ensuring alignment with **Pipecat's architecture, coding standards, and example patterns**.
+It focuses on **readability, correctness, performance, and consistency**, while avoiding breaking changes.
+
+---
+
+## Skill Overview
+
+This skill analyzes all changes introduced in your branch and performs the following actions:
+
+1. **Analyze Branch Changes**
+   - Review uncommitted changes and outgoing commits
+2. **Refactor for Readability**
+   - Improve clarity, naming, structure, and modern Python usage
+3. **Enhance Performance**
+   - Identify safe, conservative optimization opportunities
+4. **Add Documentation**
+   - Apply Pipecat-style, Google-format docstrings
+5. **Ensure Pattern Consistency**
+   - Match existing Pipecat services, pipelines, and examples
+6. **Validate Examples**
+   - Ensure examples follow foundational patterns (e.g. `07-interruptible.py`)
+
+---
+
+## Usage
+
+Invoke the skill using any of the following commands:
+
+- "Clean up my branch code"
+- "Refactor the changes in my branch"
+- "Review and improve my branch code"
+- `/cleanup`
+
+---
+
+## What This Skill Does
+
+### 1. Analyze Branch Changes
+
+The skill retrieves all uncommitted changes and outgoing commits to understand:
+
+- New files added
+- Modified files
+- Code additions and deletions
+- Overall scope and intent of changes
+
+---
+
+### 2. Code Refactoring
+
+#### Readability Improvements
+
+- Replace tuples with named classes or dataclasses
+- Improve variable, method, and class naming
+- Extract complex logic into well-named helper methods
+- Add missing type hints
+- Simplify nested or complex conditionals
+- Replace deprecated methods and features
+- Normalize formatting to match Pipecat style
+
+#### Performance Enhancements
+
+- Identify inefficient loops or repeated work
+- Suggest appropriate data structures
+- Optimize async workflows and I/O
+- Remove redundant operations
+
+> Performance changes are conservative and non-breaking.
+
+---
+
+### 3. Documentation
+
+Documentation follows **Google-style docstrings**, consistent with Pipecat conventions.
+
+#### Class Documentation
+
+```python
+class ExampleService:
+    """Brief one-line description.
+
+    Detailed explanation of the class purpose, responsibilities,
+    and important behaviors.
+
+    Supported features:
+
+    - Feature 1
+    - Feature 2
+    - Feature 3
+    """
+```
+
+#### Method Documentation
+
+```python
+def process_data(self, data: str, options: Optional[dict] = None) -> bool:
+    """Process incoming data with optional configuration.
+
+    Args:
+        data: The input data to process.
+        options: Optional configuration dictionary.
+
+    Returns:
+        True if processing succeeded, False otherwise.
+
+    Raises:
+        ValueError: If data is empty or invalid.
+    """
+```
+
+#### Pydantic Model Parameters
+
+```python
+class InputParams(BaseModel):
+    """Configuration parameters for the service.
+
+    Parameters:
+        timeout: Request timeout in seconds.
+        retry_count: Number of retry attempts.
+        enable_logging: Whether to enable debug logging.
+    """
+
+    timeout: Optional[float] = None
+    retry_count: int = 3
+    enable_logging: bool = False
+```
+
+---
+
+### 4. Pattern Consistency Checks
+
+#### Service Classes
+
+- Correct inheritance (`TTSService`, `STTService`, `LLMService`)
+- Consistent constructor signatures
+- Frame emission patterns
+- Metrics support:
+  - `can_generate_metrics()`
+  - TTFB metrics
+  - Usage metrics
+- Alignment with similar existing services
+
+#### Examples
+
+Validated against `examples/07-interruptible.py`:
+
+- Proper `create_transport()` usage
+- Correct pipeline structure
+- Task setup and observers
+- Event handler registration
+- Runner and bot entrypoint consistency
+
+---
+
+### 5. Specific Implementation Patterns
+
+#### Service Implementation
+
+```python
+class ExampleTTSService(TTSService):
+
+    def __init__(self, *, api_key: Optional[str] = None, **kwargs):
+        super().__init__(**kwargs)
+        self._api_key = api_key or os.getenv("SERVICE_API_KEY")
+
+    def can_generate_metrics(self) -> bool:
+        return True
+
+    async def run_tts(self, text: str) -> AsyncGenerator[Frame, None]:
+        try:
+            await self.start_ttfb_metrics()
+            yield TTSStartedFrame()
+            # ... processing ...
+            yield TTSAudioRawFrame(...)
+        finally:
+            await self.stop_ttfb_metrics()
+```
+
+---
+
+#### Example Structure Pattern
+
+```python
+transport_params = {
+    "daily": lambda: DailyParams(...),
+    "twilio": lambda: FastAPIWebsocketParams(...),
+    "webrtc": lambda: TransportParams(...),
+}
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    stt = DeepgramSTTService(...)
+    tts = SomeTTSService(...)
+    llm = OpenAILLMService(...)
+
+    context = LLMContext(messages)
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(...)
+
+    pipeline = Pipeline([...])
+    task = PipelineTask(pipeline, params=..., observers=[...])
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        await task.queue_frames([LLMRunFrame()])
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+    await runner.run(task)
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+```
+
+---
+
+## Execution Flow
+
+1. Fetch uncommitted and outgoing changes
+2. Categorize files (services, examples, tests, utilities)
+3. Analyze each file:
+   - Readability
+   - Performance
+   - Documentation
+   - Pattern consistency
+4. Generate actionable recommendations
+5. Apply Pipecat standards
+
+---
+
+## Examples
+
+### Before: Tuple Usage
+
+```python
+def get_audio_info(self) -> Tuple[int, int]:
+    return (48000, 1)
+```
+
+### After: Named Class
+
+```python
+class AudioInfo:
+    """Audio configuration information.
+
+    Parameters:
+        sample_rate: Sample rate in Hz.
+        num_channels: Number of audio channels.
+    """
+
+    sample_rate: int
+    num_channels: int
+
+def get_audio_info(self) -> AudioInfo:
+    return AudioInfo(sample_rate=48000, num_channels=1)
+```
+
+---
+
+### Before: Missing Documentation
+
+```python
+class NewTTSService(TTSService):
+    def __init__(self, api_key: str, voice: str):
+        self._api_key = api_key
+        self._voice = voice
+```
+
+### After: Fully Documented
+
+```python
+class NewTTSService(TTSService):
+    """Text-to-speech service using NewProvider API.
+
+    Streams PCM audio and emits TTSAudioRawFrame frames compatible
+    with Pipecat transports.
+
+    Supported features:
+    - Text-to-speech synthesis
+    - Streaming PCM audio
+    - Voice customization
+    - TTFB metrics
+    """
+
+    def __init__(self, *, api_key: str, voice: str, **kwargs):
+        """Initialize the NewTTSService.
+
+        Args:
+            api_key: API key for authentication.
+            voice: Voice identifier to use.
+            **kwargs: Additional arguments passed to the parent service.
+        """
+        super().__init__(**kwargs)
+        self._api_key = api_key
+        self.set_voice(voice)
+```
+
+---
+
+## Notes
+
+- Non-breaking improvements only
+- Backward compatibility preserved
+- Conservative performance changes
+- Google-style docstrings
+- Pattern checks follow recent Pipecat code

.claude/skills/code-review/SKILL.md

@@ -0,0 +1,107 @@
+---
+name: code-review
+description: Automated code review for pull requests using multiple specialized agents
+disable-model-invocation: true
+allowed-tools: Bash(gh issue view:*), Bash(gh search:*), Bash(gh issue list:*), Bash(gh pr comment:*), Bash(gh pr diff:*), Bash(gh pr view:*), Bash(gh pr list:*)
+---
+
+Provide a code review for the given pull request.
+
+**Agent assumptions (applies to all agents and subagents):**
+
+- All tools are functional and will work without error. Do not test tools or make exploratory calls. Make sure this is clear to every subagent that is launched.
+- Only call a tool if it is required to complete the task. Every tool call should have a clear purpose.
+
+To do this, follow these steps precisely:
+
+1. Launch a haiku agent to check if any of the following are true:
+   - The pull request is closed
+   - The pull request is a draft
+   - The pull request does not need code review (e.g. automated PR, trivial change that is obviously correct)
+   - Claude has already commented on this PR (check `gh pr view <PR> --comments` for comments left by claude)
+
+   If any condition is true, stop and do not proceed.
+
+Note: Still review Claude generated PR's.
+
+2. Launch a haiku agent to return a list of file paths (not their contents) for all relevant CLAUDE.md files including:
+   - The root CLAUDE.md file, if it exists
+   - Any CLAUDE.md files in directories containing files modified by the pull request
+
+3. Launch a sonnet agent to view the pull request and return a summary of the changes
+
+4. Launch 4 agents in parallel to independently review the changes. Each agent should return the list of issues, where each issue includes a description and the reason it was flagged (e.g. "CLAUDE.md adherence", "bug"). The agents should do the following:
+
+   Agents 1 + 2: CLAUDE.md compliance sonnet agents
+   Audit changes for CLAUDE.md compliance in parallel. Note: When evaluating CLAUDE.md compliance for a file, you should only consider CLAUDE.md files that share a file path with the file or parents.
+
+   Agent 3: Opus bug agent (parallel subagent with agent 4)
+   Scan for obvious bugs. Focus only on the diff itself without reading extra context. Flag only significant bugs; ignore nitpicks and likely false positives. Do not flag issues that you cannot validate without looking at context outside of the git diff.
+
+   Agent 4: Opus bug agent (parallel subagent with agent 3)
+   Look for problems that exist in the introduced code. This could be security issues, incorrect logic, etc. Only look for issues that fall within the changed code.
+
+   **CRITICAL: We only want HIGH SIGNAL issues.** Flag issues where:
+   - The code will fail to compile or parse (syntax errors, type errors, missing imports, unresolved references)
+   - The code will definitely produce wrong results regardless of inputs (clear logic errors)
+   - Clear, unambiguous CLAUDE.md violations where you can quote the exact rule being broken
+
+   Do NOT flag:
+   - Code style or quality concerns
+   - Potential issues that depend on specific inputs or state
+   - Subjective suggestions or improvements
+
+   If you are not certain an issue is real, do not flag it. False positives erode trust and waste reviewer time.
+
+   In addition to the above, each subagent should be told the PR title and description. This will help provide context regarding the author's intent.
+
+5. For each issue found in the previous step by agents 3 and 4, launch parallel subagents to validate the issue. These subagents should get the PR title and description along with a description of the issue. The agent's job is to review the issue to validate that the stated issue is truly an issue with high confidence. For example, if an issue such as "variable is not defined" was flagged, the subagent's job would be to validate that is actually true in the code. Another example would be CLAUDE.md issues. The agent should validate that the CLAUDE.md rule that was violated is scoped for this file and is actually violated. Use Opus subagents for bugs and logic issues, and sonnet agents for CLAUDE.md violations.
+
+6. Filter out any issues that were not validated in step 5. This step will give us our list of high signal issues for our review.
+
+7. If issues were found, skip to step 8 to post comments.
+
+   If NO issues were found, post a summary comment using `gh pr comment` (if `--comment` argument is provided):
+   "No issues found. Checked for bugs and CLAUDE.md compliance."
+
+8. Create a list of all comments that you plan on leaving. This is only for you to make sure you are comfortable with the comments. Do not post this list anywhere.
+
+9. Post inline comments for each issue using `gh pr review` with inline comments. For each comment:
+   - Provide a brief description of the issue
+   - For small, self-contained fixes, include a committable suggestion block
+   - For larger fixes (6+ lines, structural changes, or changes spanning multiple locations), describe the issue and suggested fix without a suggestion block
+   - Never post a committable suggestion UNLESS committing the suggestion fixes the issue entirely. If follow up steps are required, do not leave a committable suggestion.
+
+   **IMPORTANT: Only post ONE comment per unique issue. Do not post duplicate comments.**
+
+Use this list when evaluating issues in Steps 4 and 5 (these are false positives, do NOT flag):
+
+- Pre-existing issues
+- Something that appears to be a bug but is actually correct
+- Pedantic nitpicks that a senior engineer would not flag
+- Issues that a linter will catch (do not run the linter to verify)
+- General code quality concerns (e.g., lack of test coverage, general security issues) unless explicitly required in CLAUDE.md
+- Issues mentioned in CLAUDE.md but explicitly silenced in the code (e.g., via a lint ignore comment)
+
+Notes:
+
+- Use gh CLI to interact with GitHub (e.g., fetch pull requests, create comments). Do not use web fetch.
+- Create a todo list before starting.
+- You must cite and link each issue in inline comments (e.g., if referring to a CLAUDE.md, include a link to it).
+- If no issues are found, post a comment with the following format:
+
+---
+
+## Code review
+
+No issues found. Checked for bugs and CLAUDE.md compliance.
+
+---
+
+- When linking to code in inline comments, follow the following format precisely, otherwise the Markdown preview won't render correctly: `https://github.com/OWNER/REPO/blob/FULL_SHA/path/to/file.py#L10-L15`
+  - Requires full git sha
+  - You must provide the full sha. Commands like `https://github.com/owner/repo/blob/$(git rev-parse HEAD)/foo/bar` will not work, since your comment will be directly rendered in Markdown.
+  - Repo name must match the repo you're code reviewing
+  - # sign after the file name
+  - Line range format is L[start]-L[end]
+  - Provide at least 1 line of context before and after, centered on the line you are commenting about (eg. if you are commenting about lines 5-6, you should link to `L4-7`)

.claude/skills/docstring/SKILL.md

@@ -0,0 +1,256 @@
+---
+name: docstring
+description: Document a Python module and its classes using Google style
+---
+
+Document a Python module or class using Google-style docstrings following project conventions. The argument can be a class name or a module path.
+
+## Instructions
+
+1. Determine what to document based on the argument:
+
+   **If a module path is provided** (e.g. `src/pipecat/audio/vad/vad_analyzer.py`):
+   - Use that file directly
+
+   **If a class name is provided** (e.g. `VADAnalyzer`):
+   - Search for `class ClassName` in `src/pipecat/`
+   - If multiple files contain that class name, list all matches with their file paths, ask the user which one they want to document, and wait for confirmation
+
+2. Once the file is identified, read the module to understand its structure:
+   - Identify all classes, functions, and important type aliases
+   - Understand the purpose of each component
+
+4. Apply documentation in this order:
+   - Module docstring (at top, after imports)
+   - Class docstrings
+   - `__init__` methods (always document constructor parameters)
+   - Public methods (not starting with `_`)
+   - Dataclass/config classes with field descriptions
+
+5. Skip documentation for:
+   - Private methods (starting with `_`)
+   - Simple dunder methods (`__str__`, `__repr__`, `__post_init__`)
+   - Very simple pass-through properties
+   - **Already documented code** - If a class, method, or function already has a complete docstring that follows the project style, do not modify it. A docstring is complete if it has:
+     - A one-line summary
+     - Args section (if it has parameters)
+     - Returns section (if it returns something meaningful)
+   - Only add or improve documentation where it is missing or incomplete
+
+## Module Docstring Format
+
+```python
+"""[One-line description of module purpose].
+
+[Optional: Longer explanation of functionality, key classes, or use cases.]
+"""
+```
+
+Example:
+```python
+"""Neuphonic text-to-speech service implementations.
+
+This module provides WebSocket and HTTP-based integrations with Neuphonic's
+text-to-speech API for real-time audio synthesis.
+"""
+```
+
+## Class Docstring Format
+
+```python
+class ClassName:
+    """One-line summary describing what the class does.
+
+    [Longer description explaining purpose, behavior, and key features.
+    Use action-oriented language.]
+
+    [Optional: Event handlers, usage notes, or important caveats.]
+    """
+```
+
+Example:
+```python
+class FrameProcessor(BaseObject):
+    """Base class for all frame processors in the pipeline.
+
+    Frame processors are the building blocks of Pipecat pipelines, they can be
+    linked to form complex processing pipelines. They receive frames, process
+    them, and pass them to the next or previous processor in the chain.
+
+    Event handlers available:
+
+    - on_before_process_frame: Called before a frame is processed
+    - on_after_process_frame: Called after a frame is processed
+
+    Example::
+
+        @processor.event_handler("on_before_process_frame")
+        async def on_before_process_frame(processor, frame):
+            ...
+
+        @processor.event_handler("on_after_process_frame")
+        async def on_after_process_frame(processor, frame):
+            ...
+    """
+```
+
+Note: When listing event handlers, do NOT use backticks. Include an `Example::` section (with double colon for Sphinx) showing the decorator pattern and function signature for each event.
+
+## Constructor (`__init__`) Format
+
+```python
+def __init__(self, *, param1: Type, param2: Type = default, **kwargs):
+    """Initialize the [ClassName].
+
+    Args:
+        param1: Description of param1 and its purpose.
+        param2: Description of param2. Defaults to [default].
+        **kwargs: Additional arguments passed to parent class.
+    """
+```
+
+Example:
+```python
+def __init__(
+    self,
+    *,
+    api_key: str,
+    voice_id: Optional[str] = None,
+    sample_rate: Optional[int] = 22050,
+    **kwargs,
+):
+    """Initialize the Neuphonic TTS service.
+
+    Args:
+        api_key: Neuphonic API key for authentication.
+        voice_id: ID of the voice to use for synthesis.
+        sample_rate: Audio sample rate in Hz. Defaults to 22050.
+        **kwargs: Additional arguments passed to parent InterruptibleTTSService.
+    """
+```
+
+## Method Docstring Format
+
+```python
+async def method_name(self, param1: Type) -> ReturnType:
+    """One-line summary of what method does.
+
+    [Longer description if behavior isn't obvious.]
+
+    Args:
+        param1: Description of param1.
+
+    Returns:
+        Description of return value.
+
+    Raises:
+        ExceptionType: When this exception is raised.
+    """
+```
+
+Example:
+```python
+async def put(self, item: Tuple[Frame, FrameDirection, FrameCallback]):
+    """Put an item into the priority queue.
+
+    System frames (`SystemFrame`) have higher priority than any other
+    frames. If a non-frame item is provided it will have the highest priority.
+
+    Args:
+        item: The item to enqueue.
+    """
+```
+
+## Dataclass/Config Format
+
+```python
+@dataclass
+class ConfigName:
+    """One-line description of configuration.
+
+    [Explanation of when/how to use this config.]
+
+    Parameters:
+        field1: Description of field1.
+        field2: Description of field2. Defaults to [default].
+    """
+
+    field1: Type
+    field2: Type = default_value
+```
+
+Example:
+```python
+@dataclass
+class FrameProcessorSetup:
+    """Configuration parameters for frame processor initialization.
+
+    Parameters:
+        clock: The clock instance for timing operations.
+        task_manager: The task manager for handling async operations.
+        observer: Optional observer for monitoring frame processing events.
+    """
+
+    clock: BaseClock
+    task_manager: BaseTaskManager
+    observer: Optional[BaseObserver] = None
+```
+
+## Enum Documentation Format
+
+```python
+class EnumName(Enum):
+    """One-line description of the enum purpose.
+
+    [Longer description of how the enum is used.]
+
+    Parameters:
+        VALUE1: Description of VALUE1.
+        VALUE2: Description of VALUE2.
+    """
+
+    VALUE1 = 1
+    VALUE2 = 2
+```
+
+## Writing Style Guidelines
+
+- **Concise and professional** - No casual language or filler words
+- **Action-oriented** - Start with verbs: "Processes...", "Manages...", "Converts..."
+- **Purpose before implementation** - Explain WHY before HOW
+- **Clear parameter descriptions** - Include type hints, defaults, and purpose
+- **No redundant type info** - Type hints are in the signature, don't repeat in description
+- **Use backticks for code references** - Wrap class names, method names, event names, parameter names, and code snippets in backticks
+
+Good: "Neuphonic API key for authentication."
+Bad: "str: The API key (string) that is used for authenticating with Neuphonic."
+
+Good: "Triggers `on_speech_started` when the `VADAnalyzer` detects speech."
+Bad: "Triggers on_speech_started when the VADAnalyzer detects speech."
+
+## Deprecation Notice Format
+
+When documenting deprecated code:
+
+```python
+"""[Description].
+
+.. deprecated:: X.X.X
+    `ClassName` is deprecated and will be removed in a future version.
+    Use `NewClassName` instead.
+"""
+```
+
+## Checklist
+
+Before finishing, verify:
+
+- [ ] Module has a docstring at the top (after copyright header and imports)
+- [ ] All public classes have docstrings
+- [ ] All `__init__` methods document their parameters
+- [ ] All public methods have docstrings with Args/Returns/Raises as needed
+- [ ] Dataclasses use "Parameters:" section for field descriptions
+- [ ] Enums document each value in "Parameters:" section
+- [ ] Writing is concise and action-oriented
+- [ ] No documentation added to private methods (starting with `_`)
+- [ ] Existing complete docstrings were left unchanged

.claude/skills/pr-description/SKILL.md

@@ -0,0 +1,128 @@
+---
+name: pr-description
+description: Update a GitHub PR description with a summary of changes
+---
+
+Update a GitHub pull request description based on the changes in the PR.
+
+## Arguments
+
+```
+/pr-description <PR_NUMBER> [--fixes <ISSUE_NUMBERS>]
+```
+
+- `PR_NUMBER` (required): The pull request number to update
+- `--fixes` (optional): Comma-separated issue numbers that this PR fixes (e.g., `--fixes 123,456`)
+
+Examples:
+- `/pr-description 3534`
+- `/pr-description 3534 --fixes 123`
+- `/pr-description 3534 --fixes 123,456,789`
+
+## Instructions
+
+1. First, gather information about the PR:
+   - Use GitHub plugin to get PR details (title, current description, base branch)
+   - Use local git to get commits: `git log main..HEAD --oneline`
+   - Use local git to get the diff: `git diff main..HEAD`
+   - Parse any `--fixes` argument for issue numbers
+
+2. Check the existing PR description:
+   - If it already has a complete, accurate description that reflects the changes, do nothing
+   - If it's missing sections, incomplete, or outdated compared to the actual changes, proceed to update
+   - If it only has the template placeholder text, generate a full description
+
+3. Analyze the changes:
+   - Understand the purpose of each commit
+   - Identify any breaking changes (API changes, removed features, behavior changes)
+   - Look for new features, bug fixes, refactoring, or documentation changes
+   - Collect issue numbers from:
+     - The `--fixes` argument (if provided)
+     - Commit messages (patterns like "Fixes #123", "Closes #456", "Resolves #789")
+
+4. Generate or update the PR description with these sections:
+
+## PR Description Format
+
+### Summary (always include)
+
+Brief bullet points describing what changed and why. Focus on the *purpose* and *impact*, not implementation details.
+
+```markdown
+## Summary
+
+- Added X to enable Y
+- Fixed bug where Z would happen
+- Refactored W for better maintainability
+```
+
+### Breaking Changes (include only if applicable)
+
+Document any changes that affect existing users or APIs.
+
+```markdown
+## Breaking Changes
+
+- `ClassName.method()` now requires a `param` argument
+- Removed deprecated `old_function()` - use `new_function()` instead
+```
+
+### Testing (include when non-obvious)
+
+How to verify the changes work. Skip for trivial changes.
+
+```markdown
+## Testing
+
+- Run `uv run pytest tests/test_feature.py` to verify the fix
+- Example usage: `uv run examples/new_feature.py`
+```
+
+### Fixes (include if issues are provided or found in commits)
+
+List issues this PR fixes. GitHub will automatically close these issues when the PR is merged.
+
+```markdown
+## Fixes
+
+- Fixes #123
+- Fixes #456
+```
+
+Note: Use "Fixes #X" format (not "Closes" or "Resolves") for consistency. Each issue should be on its own line with "Fixes" to ensure GitHub auto-closes them.
+
+## Guidelines
+
+- **Be concise** - Reviewers should understand the PR in 30 seconds
+- **Focus on why** - The diff shows *what* changed, explain *why*
+- **Skip empty sections** - Only include sections that have content
+- **Use bullet points** - Easier to scan than paragraphs
+- **Don't duplicate the diff** - Avoid listing every file or line changed
+
+## Example Output
+
+```markdown
+## Summary
+
+- Added `/docstring` skill for documenting Python modules with Google-style docstrings
+- Skill finds classes by name and handles conflicts when multiple matches exist
+- Skips already-documented code to avoid unnecessary changes
+
+## Testing
+
+/docstring ClassName
+
+## Fixes
+
+- Fixes #123
+```
+
+## Checklist
+
+Before updating the PR:
+
+- [ ] Verified existing description needs updating (not already complete)
+- [ ] Summary accurately reflects the changes
+- [ ] Breaking changes are clearly documented (if any)
+- [ ] No unnecessary sections included
+- [ ] Description is concise and scannable

.claude/skills/pr-submit/SKILL.md

@@ -0,0 +1,28 @@
+---
+name: pr-submit
+description: Create and submit a GitHub PR from the current branch
+---
+
+Submit the current changes as a GitHub pull request.
+
+## Instructions
+
+1. Check the current state of the repository:
+   - Run `git status` to see staged, unstaged, and untracked changes
+   - Run `git diff` to see current changes
+   - Run `git log --oneline -10` to see recent commits
+
+2. If there are uncommitted changes relevant to the PR:
+   - Ask the user if they want a specific prefix for the branch name (e.g., `alice/`, `fix/`, `feat/`)
+   - Create a new branch based on the current branch
+   - Commit the changes using multiple commits if the changes are unrelated
+
+3. Push the branch and create the PR:
+   - Push with `-u` flag to set upstream tracking
+   - Create the PR using `gh pr create`
+
+4. After the PR is created:
+   - Run `/changelog <pr_number>` to generate changelog files, then commit and push them
+   - Run `/pr-description <pr_number>` to update the PR description
+
+5. Return the PR URL to the user.

.claude/skills/squash-commits/SKILL.md

@@ -0,0 +1,91 @@
+---
+name: squash-commits
+description: Reorganize messy branch commits into a small set of logical, meaningful commits without changing any content. Drops merge-from-main commits. Safe: creates a backup branch first.
+---
+
+Reorganize the commits on the current branch into a small number of logical commits. Do NOT change any file content — only the commit structure changes.
+
+## Instructions
+
+### 1. Safety check
+
+```bash
+git status --short
+```
+
+If there are uncommitted changes, stop and tell the user to commit or stash them first.
+
+### 2. Inspect the branch
+
+```bash
+git log main..HEAD --oneline
+git diff main..HEAD --name-only
+```
+
+List every file changed vs `main` and every commit on the branch (excluding merge commits from main).
+
+### 3. Create a backup branch
+
+```bash
+git branch backup/<current-branch-name>
+```
+
+Tell the user the backup exists so they can recover if needed.
+
+### 4. Soft-reset to main and unstage everything
+
+```bash
+git reset --soft main
+git restore --staged .
+```
+
+All branch changes are now in the working tree, unstaged. No content has changed.
+
+### 5. Plan the logical groups
+
+Read the changed files and the original commit messages to understand what the work covers. Group related files into logical commits. Typical groups:
+
+- Core feature or fix (new source files + modified core files)
+- Secondary features or fixes (each as its own commit if distinct)
+- Refactoring or renames
+- Tests
+- Changelogs / docs
+
+Use the changelog files (if any) as a strong hint — each changelog entry often maps to one commit.
+
+Present the proposed grouping to the user and ask for confirmation before committing.
+
+### 6. Commit in logical groups
+
+For each group, stage only the relevant files and commit with a clear message following the project's conventions:
+
+```bash
+git add <file1> <file2> ...
+git commit -m "..."
+```
+
+Use conventional commit prefixes if the project uses them (`feat:`, `fix:`, `refactor:`, `test:`, `chore:`).
+
+### 7. Verify
+
+```bash
+git log main..HEAD --oneline
+git diff main..HEAD --name-only
+git status --short
+```
+
+Confirm:
+- Commit count is small and each message is meaningful
+- The set of changed files vs `main` is identical to before
+- Working tree is clean
+
+### 8. Remind about force-push
+
+The branch history has been rewritten. Tell the user they will need to `git push --force-with-lease` when they are ready to update the remote. Do NOT push automatically.
+
+## Rules
+
+- Never change file contents. If you find yourself editing a file, stop.
+- Never skip the backup branch step.
+- Never force-push without explicit user instruction.
+- If any step fails or the result looks wrong, tell the user and suggest restoring from the backup: `git reset --hard backup/<branch-name>`.

.claude/skills/update-docs/SKILL.md

@@ -0,0 +1,306 @@
+---
+name: update-docs
+description: Update documentation pages to match source code changes on the current branch
+---
+
+Update documentation pages to reflect source code changes on the current branch. Analyzes the diff against main, maps changed source files to their corresponding doc pages, and makes targeted edits.
+
+## Arguments
+
+```
+/update-docs [DOCS_PATH]
+```
+
+- `DOCS_PATH` (optional): Path to the docs repository root. If not provided, ask the user.
+
+Examples:
+- `/update-docs /Users/me/src/docs`
+- `/update-docs`
+
+## Instructions
+
+### Step 1: Resolve docs path
+
+If `DOCS_PATH` was provided as an argument, use it. Otherwise, ask the user for the path to their docs repository.
+
+Verify the path exists and contains `server/services/` subdirectory.
+
+### Step 2: Create docs branch
+
+Get the current pipecat branch name:
+```bash
+git rev-parse --abbrev-ref HEAD
+```
+
+In the docs repo, create a new branch off main with a matching name:
+```bash
+cd DOCS_PATH && git checkout main && git pull && git checkout -b {branch-name}-docs
+```
+
+For example, if the pipecat branch is `feat/new-service`, the docs branch becomes `feat/new-service-docs`.
+
+All doc edits in subsequent steps are made on this branch.
+
+### Step 3: Detect changed source files
+
+Run:
+```bash
+git diff main..HEAD --name-only
+```
+
+Filter to files that could affect documentation:
+- `src/pipecat/services/**/*.py` (service implementations)
+- `src/pipecat/transports/**/*.py` (transport implementations)
+- `src/pipecat/serializers/**/*.py` (serializer implementations)
+- `src/pipecat/processors/**/*.py` (processor implementations)
+- `src/pipecat/audio/**/*.py` (audio utilities)
+- `src/pipecat/turns/**/*.py` (turn management)
+- `src/pipecat/observers/**/*.py` (observers)
+- `src/pipecat/pipeline/**/*.py` (pipeline core)
+
+Ignore `__init__.py`, `__pycache__`, test files, and files that only contain type re-exports.
+
+### Step 4: Map source files to doc pages
+
+For each changed source file, find the corresponding doc page. Read the mapping file at `.claude/skills/update-docs/SOURCE_DOC_MAPPING.md` and apply its tiered lookup: tier 1 (known exceptions) → tier 2 (pattern matching) → tier 3 (search fallback). **First match wins.**
+
+### Step 5: Analyze each source-doc pair
+
+For each mapped pair:
+
+1. **Read the full source file** to understand current state
+2. **Read the diff** for that file: `git diff main..HEAD -- <source_file>`
+3. **Read the current doc page** in full
+
+Identify what changed by comparing source to docs:
+
+- **Constructor parameters**: Compare `__init__` signature to the Configuration section's `<ParamField>` entries
+- **InputParams fields**: Compare `InputParams(BaseModel)` class fields to the InputParams table
+- **Event handlers**: Compare `_register_event_handler` calls and event handler definitions to Event Handlers section
+- **Class names / imports**: Check if Usage examples reference correct names
+- **Behavioral changes**: Check if Notes section needs updating
+
+### Step 6: Make targeted edits
+
+For each doc page that needs updates, edit **only the sections that need changes**. Preserve all other content exactly as-is.
+
+#### Rules
+
+- **Never remove content** unless the corresponding source code was removed
+- **Never rewrite sections** that are already accurate
+- **Match existing formatting** — if the page uses `<ParamField>` tags, use them; if it uses tables, use tables
+- **Keep descriptions concise** — match the tone and length of surrounding content
+- **Preserve CardGroup, links, and examples** unless they reference removed functionality
+- **Don't touch frontmatter** unless the class was renamed
+
+#### Section-specific guidance
+
+**Configuration** (constructor params):
+- Use `<ParamField path="name" type="type" default="value">` format if the page already uses it
+- Add new params in logical order (required first, then optional)
+- Remove params that no longer exist in source
+- Update types/defaults that changed
+
+**InputParams** (runtime settings):
+- Use markdown table format: `| Parameter | Type | Default | Description |`
+- Match the field names and types from the `InputParams(BaseModel)` class
+- Include the default values from the source
+
+**Usage** (code examples):
+- Update import paths, class names, and parameter names
+- Only modify examples if they would break or be misleading with the new API
+- Don't rewrite working examples just to add new optional params
+
+**Notes**:
+- Add notes for new behavioral gotchas or breaking changes
+- Remove notes about limitations that were fixed
+- Keep existing notes that are still accurate
+
+**Event Handlers**:
+- Update the event table and example code
+- Add new events, remove deleted ones
+- Update handler signatures if they changed
+
+**Overview / Key Features / Prerequisites**:
+- Only update if the PR fundamentally changes what the service does (new capability, removed capability, renamed class)
+- Most PRs will NOT need changes to these sections
+
+### Step 7: Update guides
+
+Guides at `DOCS_PATH/guides/` reference specific class names, parameters, imports, and code patterns. After completing reference doc edits, check if any guides need updates too.
+
+For each changed source file, collect the class names, renamed parameters, and changed imports from the diff. Search the guides directory:
+```bash
+grep -rl "ClassName\|old_param_name" DOCS_PATH/guides/
+```
+
+For each guide that references changed code:
+1. Read the full guide
+2. Update class names, parameter names, import paths, and code examples that are now incorrect
+3. **Don't rewrite prose** — only fix the specific references that changed
+4. Leave guides alone if they reference the service generally but don't use any changed APIs
+
+Guide directories:
+- `guides/learn/` — conceptual tutorials (pipeline, LLM, STT, TTS, etc.)
+- `guides/fundamentals/` — practical how-tos (metrics, recording, transcripts, etc.)
+- `guides/features/` — feature-specific guides (Gemini Live, OpenAI audio, WhatsApp, etc.)
+- `guides/telephony/` — telephony integration guides (Twilio, Plivo, Telnyx, etc.)
+
+### Step 8: Identify doc gaps
+
+After processing all mapped pairs, check for two kinds of gaps:
+
+**Missing pages**: Source files that had no doc page mapping (neither tier 1, 2, nor 3) and are not marked as "(skip)". For each, tell the user:
+- The source file path
+- The main class(es) it defines
+- Whether a new doc page should be created
+
+**Missing sections**: Mapped doc pages that are missing standard sections compared to the source. For example, a transport page with no Configuration section, or a service page with no InputParams table when the source defines `InputParams(BaseModel)`. Flag these and offer to add the missing sections.
+
+If the user wants a new page, do all three of the following:
+
+#### 8a: Create the doc page
+
+Create the new `.mdx` file using this template structure:
+```
+---
+title: "Service Name"
+description: "Brief description"
+---
+
+## Overview
+
+[Description from class docstring or source analysis]
+
+<CardGroup cols={2}>
+  [Cards for API reference and examples if available]
+</CardGroup>
+
+## Installation
+
+```bash
+pip install "pipecat-ai[package-name]"
+```
+
+## Prerequisites
+
+[Environment variables and account setup]
+
+## Configuration
+
+[ParamField entries for constructor params]
+
+## InputParams
+
+[Table of InputParams fields, if the service has them]
+
+## Usage
+
+### Basic Setup
+
+```python
+[Minimal working example]
+```
+
+## Notes
+
+[Important caveats]
+
+## Event Handlers
+
+[Event table and example code]
+```
+
+#### 8b: Add to docs.json
+
+Add the new page path to `DOCS_PATH/docs.json` in the correct navigation group. The path format is `server/services/{category}/{provider}` (without the `.mdx` extension).
+
+Find the matching group in the navigation structure:
+- **STT** → `"group": "Speech-to-Text"` under Services
+- **TTS** → `"group": "Text-to-Speech"` under Services
+- **LLM** → `"group": "LLM"` under Services
+- **S2S** → `"group": "Speech-to-Speech"` under Services
+- **Transport** → `"group": "Transport"` under Services
+- **Serializer** → `"group": "Serializers"` under Services
+- **Image generation** → `"group": "Image Generation"` under Services
+- **Video** → `"group": "Video"` under Services
+- **Memory** → `"group": "Memory"` under Services
+- **Vision** → `"group": "Vision"` under Services
+- **Analytics** → `"group": "Analytics & Monitoring"` under Services
+
+Insert the new entry **alphabetically** within the group's `pages` array. For example, adding a new STT service "foo":
+```json
+{
+  "group": "Speech-to-Text",
+  "pages": [
+    "server/services/stt/assemblyai",
+    "server/services/stt/aws",
+    ...
+    "server/services/stt/foo",
+    ...
+  ]
+}
+```
+
+#### 8c: Add to supported-services.mdx
+
+Add a new row to the correct category table in `DOCS_PATH/server/services/supported-services.mdx`.
+
+Use this format:
+```
+| [DisplayName](/server/services/{category}/{provider}) | `pip install "pipecat-ai[package]"` |
+```
+
+To determine the correct values:
+- **DisplayName**: Use the service's human-readable name (e.g., "ElevenLabs", "AWS Polly", "Google Gemini")
+- **package**: Look at the service's `pyproject.toml` extras or the import pattern in the source code. For example, if the service is in `src/pipecat/services/foo/`, the package is typically `foo`.
+- If no pip dependencies are required, use `No dependencies required` instead.
+
+Insert the new row **alphabetically** within the table. Match the column alignment of the existing rows.
+
+### Step 9: Output summary
+
+After all edits are complete, print a summary:
+
+```
+## Documentation Updates
+
+### Updated reference pages
+- `server/services/stt/deepgram.mdx` — Updated Configuration (added `new_param`), InputParams (updated `language` default)
+- `server/services/tts/elevenlabs.mdx` — Updated Event Handlers (added `on_connected`)
+
+### Updated guides
+- `guides/learn/speech-to-text.mdx` — Updated code example (renamed `old_param` → `new_param`)
+
+### New service pages
+- `server/services/tts/newprovider.mdx` — Created page, added to docs.json (Text-to-Speech), added to supported-services.mdx
+
+### Unmapped source files
+- `src/pipecat/services/newprovider/tts.py` — NewProviderTTSService (no doc page exists)
+
+### Skipped files
+- `src/pipecat/services/ai_service.py` — internal base class
+```
+
+## Guidelines
+
+- **Be conservative** — only change what the diff warrants. Don't "improve" docs beyond what changed in source.
+- **Read before editing** — always read the full doc page before making changes so you understand the existing structure.
+- **Preserve voice** — match the writing style of the existing doc page, don't impose a different tone.
+- **One PR at a time** — this skill operates on the current branch's diff against main. Don't look at other branches.
+- **Parallel analysis** — when multiple source files map to different doc pages, analyze and edit them in parallel for efficiency.
+- **Shared source files** — files like `services/google/google.py` are shared bases. Check which services import from them and update all affected doc pages.
+
+## Checklist
+
+Before finishing, verify:
+
+- [ ] All changed source files were checked against the mapping table
+- [ ] Each doc page edit matches the actual source code change (not guessed)
+- [ ] No content was removed unless the corresponding source was removed
+- [ ] New parameters have accurate types and defaults from source
+- [ ] Formatting matches the existing page style
+- [ ] Guides referencing changed APIs were checked and updated
+- [ ] New service pages were added to `docs.json` in the correct group, alphabetically
+- [ ] New service pages were added to `supported-services.mdx` in the correct table, alphabetically
+- [ ] Unmapped files were reported to the user

.claude/skills/update-docs/SOURCE_DOC_MAPPING.md

@@ -0,0 +1,79 @@
+# Source-to-Doc Mapping
+
+Maps pipecat source files to their documentation pages. Source paths are relative to `src/pipecat/`. Doc paths are relative to `DOCS_PATH`.
+
+## Name mismatches
+
+These source paths don't follow the standard `services/{provider}/{type}.py` → `server/services/{type}/{provider}.mdx` pattern.
+
+| Source path | Doc page |
+|---|---|
+| `services/google/llm.py` | `server/services/llm/gemini.mdx` |
+| `services/google/llm_vertex.py` | `server/services/llm/google-vertex.mdx` |
+| `services/google/google.py` | (shared base — check which services use it) |
+| `services/google/gemini_live/**` | `server/services/s2s/gemini-live.mdx` |
+| `services/google/gemini_live/llm_vertex.py` | `server/services/s2s/gemini-live-vertex.mdx` |
+| `services/aws_nova_sonic/**` | `server/services/s2s/aws.mdx` |
+| `services/ultravox/**` | `server/services/s2s/ultravox.mdx` |
+| `services/grok/realtime/**` | `server/services/s2s/grok.mdx` |
+| `services/openai/realtime/**` | `server/services/s2s/openai.mdx` |
+| `processors/frameworks/rtvi.py` | `server/frameworks/rtvi/rtvi-processor.mdx` and `server/frameworks/rtvi/rtvi-observer.mdx` |
+| `processors/transcript_processor.py` | `server/utilities/transcript-processor.mdx` |
+| `processors/user_idle_processor.py` | `server/utilities/user-idle-processor.mdx` |
+| `processors/idle_frame_processor.py` | `server/pipeline/pipeline-idle-detection.mdx` |
+| `pipeline/task.py` | `server/pipeline/pipeline-task.mdx` |
+| `pipeline/runner.py` | `server/utilities/runner/guide.mdx` |
+| `transports/base_transport.py` | `server/services/transport/transport-params.mdx` |
+
+## Skip list
+
+These files should never trigger doc updates.
+
+| Pattern | Reason |
+|---|---|
+| `services/ai_service.py` | Internal base class |
+| `services/stt_service.py` | Internal base class |
+| `services/tts_service.py` | Internal base class |
+| `services/llm_service.py` | Internal base class |
+| `services/websocket_service.py` | Internal base class |
+| `services/openai_realtime_beta/**` | Deprecated |
+| `services/openai_realtime/**` | Deprecated |
+| `services/gemini_multimodal_live/**` | Deprecated |
+| `services/aws/agent_core.py` | Internal |
+| `services/aws/sagemaker/**` | No doc page |
+| `transports/base_input.py` | Internal base class |
+| `transports/base_output.py` | Internal base class |
+| `transports/websocket/client.py` | No doc page |
+| `serializers/base_serializer.py` | Internal base class |
+| `serializers/protobuf.py` | Internal |
+| `processors/audio/**` | Internal |
+| `pipeline/pipeline.py` | Core architecture, not a service doc |
+
+## Pattern matching
+
+For files not in the tables above, apply these patterns. Convert underscores to hyphens in provider names for doc filenames.
+
+| Source pattern | Doc pattern |
+|---|---|
+| `services/{provider}/stt*.py` | `server/services/stt/{provider}.mdx` |
+| `services/{provider}/tts*.py` | `server/services/tts/{provider}.mdx` |
+| `services/{provider}/llm*.py` | `server/services/llm/{provider}.mdx` |
+| `services/{provider}/image*.py` | `server/services/image-generation/{provider}.mdx` |
+| `services/{provider}/video*.py` | `server/services/video/{provider}.mdx` |
+| `services/{provider}/realtime/**` | `server/services/s2s/{provider}.mdx` |
+| `transports/{name}/**` | `server/services/transport/{name}.mdx` |
+| `serializers/{name}.py` | `server/services/serializers/{name}.mdx` |
+| `observers/**` | `server/utilities/observers/` (match by class name) |
+| `audio/vad/**` | `server/utilities/audio/` (match by class name) |
+| `audio/filters/**` | `server/utilities/audio/` (match by class name) |
+| `audio/mixers/**` | `server/utilities/audio/` (match by class name) |
+| `processors/filters/**` | `server/utilities/filters/` (match by class name) |
+
+If the doc file doesn't exist at the resolved path, the file is **unmapped**.
+
+## Search fallback
+
+For files that don't match any table or pattern above:
+1. Extract the main class name(s) from the source file
+2. Search the docs directory for that class name: `grep -r "ClassName" DOCS_PATH/server/`
+3. If found in a doc page, use that as the mapping

.dockerignore

@@ -1,30 +0,0 @@
-# flyctl launch added from .gitignore
-**/.vscode
-**/env
-**/__pycache__
-**/*~
-**/venv
-#*#
-
-# Distribution / packaging
-**/.Python
-**/build
-**/develop-eggs
-**/dist
-**/downloads
-**/eggs
-**/.eggs
-**/lib
-**/lib64
-**/parts
-**/sdist
-**/var
-**/wheels
-**/share/python-wheels
-**/*.egg-info
-**/.installed.cfg
-**/*.egg
-**/MANIFEST
-**/.DS_Store
-**/.env
-fly.toml

.github/workflows/build.yaml

@@ -21,20 +21,20 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v4
-      
+
       - name: Install uv
         uses: astral-sh/setup-uv@v3
         with:
           version: "latest"
-      
+
       - name: Set up Python
-        run: uv python install 3.10
-      
+        run: uv python install 3.12
+
       - name: Install development dependencies
         run: uv sync --group dev
-      
+
       - name: Build project
         run: uv build
-      
+
       - name: Install project in editable mode
-        run: uv pip install --editable .
+        run: uv pip install --editable .

.github/workflows/coverage.yaml

@@ -29,11 +29,23 @@ jobs:
 
       - name: Install system packages
         run: |
+          sudo apt-get update
           sudo apt-get install -y portaudio19-dev
 
       - name: Install dependencies
         run: |
-          uv sync --group dev --extra anthropic --extra aws --extra google --extra langchain
+          uv sync --group dev \
+            --extra anthropic \
+            --extra aws \
+            --extra deepgram \
+            --extra google \
+            --extra langchain \
+            --extra livekit \
+            --extra piper \
+            --extra runner \
+            --extra sagemaker \
+            --extra tracing \
+            --extra websocket
 
       - name: Run tests with coverage
         run: |

.github/workflows/format.yaml

@@ -22,22 +22,28 @@ jobs:
     steps:
       - name: Checkout repo
         uses: actions/checkout@v4
-      
+
       - name: Install uv
         uses: astral-sh/setup-uv@v3
         with:
           version: "latest"
-      
+
       - name: Set up Python
-        run: uv python install 3.10
-      
+        run: uv python install 3.12
+
       - name: Install development dependencies
-        run: uv sync --group dev
-      
+        # `--all-extras` (matching the dev setup in README.md) so pyright can
+        # resolve types from various optional dependencies.
+        run: uv sync --group dev --all-extras --no-extra gstreamer --no-extra local
+
       - name: Ruff formatter
         id: ruff-format
         run: uv run ruff format --diff
-      
+
       - name: Ruff linter (all rules)
         id: ruff-check
-        run: uv run ruff check
+        run: uv run ruff check
+
+      - name: Type check (pyright)
+        id: pyright
+        run: uv run pyright

.github/workflows/generate-changelog.yml

@@ -0,0 +1,174 @@
+name: Generate Changelog for Release
+
+on:
+  workflow_dispatch:
+    inputs:
+      version:
+        description: "Release version (e.g., 0.0.97)"
+        required: true
+        type: string
+      date:
+        description: "Release date (YYYY-MM-DD format, defaults to today)"
+        required: false
+        type: string
+        default: ""
+
+permissions:
+  contents: write
+  pull-requests: write
+
+jobs:
+  generate-changelog:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: Install uv
+        uses: astral-sh/setup-uv@v4
+        with:
+          enable-cache: true
+
+      - name: Install dependencies
+        run: |
+          uv sync --group dev
+
+      - name: Set release date
+        id: set_date
+        run: |
+          if [ -z "${{ inputs.date }}" ]; then
+            RELEASE_DATE=$(date +%Y-%m-%d)
+            echo "Using today's date: $RELEASE_DATE"
+          else
+            RELEASE_DATE="${{ inputs.date }}"
+            echo "Using provided date: $RELEASE_DATE"
+          fi
+          echo "release_date=$RELEASE_DATE" >> $GITHUB_OUTPUT
+
+      - name: Validate inputs
+        run: |
+          # Validate version format (basic check)
+          if ! [[ "${{ inputs.version }}" =~ ^[0-9]+\.[0-9]+\.[0-9]+.*$ ]]; then
+            echo "Error: Version must be in format X.Y.Z (e.g., 0.0.97)"
+            exit 1
+          fi
+
+          # Validate date format if provided
+          if [ -n "${{ inputs.date }}" ]; then
+            if ! date -d "${{ inputs.date }}" >/dev/null 2>&1; then
+              # Try macOS date format
+              if ! date -j -f "%Y-%m-%d" "${{ inputs.date }}" >/dev/null 2>&1; then
+                echo "Error: Date must be in YYYY-MM-DD format (e.g., 2025-12-04)"
+                exit 1
+              fi
+            fi
+          fi
+
+      - name: Check for changelog fragments
+        id: check_fragments
+        run: |
+          FRAGMENT_COUNT=$(find changelog -name "*.md" ! -name "_template.md.j2" | wc -l | tr -d ' ')
+          echo "fragment_count=$FRAGMENT_COUNT" >> $GITHUB_OUTPUT
+
+          if [ "$FRAGMENT_COUNT" -eq "0" ]; then
+            echo "❌ Error: No changelog fragments found in changelog/"
+            echo ""
+            echo "Cannot create a release without changelog entries."
+            echo "Add changelog fragments to the changelog/ directory (e.g., 1234.added.md) and try again."
+            exit 1
+          fi
+
+          # Validate fragment types
+          VALID_TYPES="added changed deprecated removed fixed performance security other"
+          INVALID_FRAGMENTS=""
+
+          for file in changelog/*.md; do
+            # Skip template
+            if [[ "$file" == "changelog/_template.md.j2" ]]; then
+              continue
+            fi
+            
+            # Extract type from filename (e.g., 1234.added.md -> added)
+            filename=$(basename "$file")
+            # Handle both 1234.added.md and 1234.added.2.md patterns
+            type=$(echo "$filename" | sed -E 's/^[0-9]+\.([a-z]+)(\.[0-9]+)?\.md$/\1/')
+            
+            # Check if type is valid
+            if ! echo "$VALID_TYPES" | grep -wq "$type"; then
+              INVALID_FRAGMENTS="$INVALID_FRAGMENTS\n  - $filename (type: '$type')"
+            fi
+          done
+
+          if [ -n "$INVALID_FRAGMENTS" ]; then
+            echo "❌ Error: Invalid changelog fragment types found:"
+            echo -e "$INVALID_FRAGMENTS"
+            echo ""
+            echo "Valid types are: $VALID_TYPES"
+            echo "Example: 1234.added.md, 5678.fixed.md"
+            exit 1
+          fi
+
+          echo "✓ Found $FRAGMENT_COUNT changelog fragment(s)"
+          echo "has_fragments=true" >> $GITHUB_OUTPUT
+
+      - name: Preview changelog
+        run: |
+          echo "## Preview of changelog for version ${{ inputs.version }}"
+          echo ""
+          uv run towncrier build --draft --version "${{ inputs.version }}" --date "${{ steps.set_date.outputs.release_date }}"
+
+      - name: Build changelog
+        run: |
+          uv run towncrier build --version "${{ inputs.version }}" --date "${{ steps.set_date.outputs.release_date }}" --yes
+
+      - name: Create Pull Request
+        uses: peter-evans/create-pull-request@v7
+        with:
+          token: ${{ secrets.GITHUB_TOKEN }}
+          commit-message: "Update changelog for version ${{ inputs.version }}"
+          title: "Release ${{ inputs.version }} - Changelog Update"
+          body: |
+            ## Changelog Update for Release ${{ inputs.version }}
+
+            This PR updates the CHANGELOG.md with all changes for version **${{ inputs.version }}**.
+
+            ### Summary
+            - **Version:** ${{ inputs.version }}
+            - **Date:** ${{ steps.set_date.outputs.release_date }}
+            - **Fragments processed:** ${{ steps.check_fragments.outputs.fragment_count }}
+
+            ### What this PR does
+            - ✅ Adds new release section to CHANGELOG.md
+            - ✅ Removes processed changelog fragments
+            - ✅ Ready to merge for release
+
+            ### Next Steps
+            1. Review the changelog entries below
+            2. Make any necessary edits to CHANGELOG.md if needed
+            3. Merge this PR
+            4. Continue with your release process
+
+            ---
+
+            <details>
+            <summary>📋 Preview of changes</summary>
+
+            The changelog has been updated with entries from the following fragments:
+
+            ```bash
+            ${{ steps.check_fragments.outputs.fragment_count }} fragments processed
+            ```
+
+            </details>
+          branch: changelog-${{ inputs.version }}
+          delete-branch: true
+          labels: |
+            changelog
+            release

.github/workflows/python-compatibility.yaml

@@ -14,7 +14,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        python-version: ['3.10.18', '3.11.13', '3.12.11', '3.13.5']
+        python-version: ['3.11.15', '3.12.13', '3.13.12', '3.14.3']
 
     name: Python ${{ matrix.python-version }}
     steps:
@@ -40,20 +40,9 @@ jobs:
           uv python install ${{ matrix.python-version }}
           uv python pin ${{ matrix.python-version }}
 
-      - name: Test uv sync with all extras (Python < 3.13)
-        if: "!startsWith(matrix.python-version, '3.13.')"
+      - name: Test uv sync with all extras
         run: |
-          uv sync --group dev --all-extras --no-extra krisp
-
-      - name: Test uv sync without PyTorch extras (Python 3.13+)
-        if: startsWith(matrix.python-version, '3.13.')
-        run: |
-          uv sync --group dev --all-extras \
-            --no-extra krisp \
-            --no-extra ultravox \
-            --no-extra local-smart-turn \
-            --no-extra moondream \
-            --no-extra mlx-whisper
+          uv sync --group dev --all-extras
 
       - name: Verify installation
         run: |

.github/workflows/sync-quickstart.yaml

@@ -1,51 +0,0 @@
-name: Sync Quickstart to pipecat-quickstart repo
-
-on:
-  push:
-    branches: [main]
-    paths:
-      - 'examples/quickstart/**'
-  workflow_dispatch: # Manual trigger
-
-jobs:
-  sync-quickstart:
-    runs-on: ubuntu-latest
-    steps:
-      - name: Checkout main repo
-        uses: actions/checkout@v4
-        with:
-          fetch-depth: 0
-
-      - name: Checkout quickstart repo
-        uses: actions/checkout@v4
-        with:
-          repository: pipecat-ai/pipecat-quickstart
-          token: ${{ secrets.QUICKSTART_SYNC_TOKEN }}
-          path: quickstart-repo
-
-      - name: Sync files (excluding uv.lock and README.md)
-        run: |
-          # Copy all files except uv.lock and README.md
-          find examples/quickstart -type f \
-            -not -name "README.md" \
-            -not -name "uv.lock" \
-            -exec cp {} quickstart-repo/ \;
-
-      - name: Commit and push changes
-        run: |
-          cd quickstart-repo
-          git config user.name "GitHub Action"
-          git config user.email "action@github.com"
-          git add .
-
-          # Only commit if there are changes
-          if ! git diff --staged --quiet; then
-            git commit -m "Sync from pipecat main repo
-            
-            Updated files from examples/quickstart/
-            Commit: ${{ github.sha }}
-            "
-            git push
-          else
-            echo "No changes to sync"
-          fi

.github/workflows/tests.yaml

@@ -33,11 +33,23 @@ jobs:
 
       - name: Install system packages
         run: |
+          sudo apt-get update
           sudo apt-get install -y portaudio19-dev
 
       - name: Install dependencies
         run: |
-          uv sync --group dev --extra anthropic --extra aws --extra google --extra langchain
+          uv sync --group dev \
+            --extra anthropic \
+            --extra aws \
+            --extra deepgram \
+            --extra google \
+            --extra langchain \
+            --extra livekit \
+            --extra piper \
+            --extra runner \
+            --extra sagemaker \
+            --extra tracing \
+            --extra websocket
 
       - name: Test with pytest
         run: |

.github/workflows/update-docs.yml

@@ -0,0 +1,148 @@
+name: Update Documentation on PR Merge
+
+on:
+  pull_request_target:
+    types: [closed]
+    branches: [main]
+    paths:
+      - "src/pipecat/services/**"
+      - "src/pipecat/transports/**"
+      - "src/pipecat/serializers/**"
+      - "src/pipecat/processors/**"
+      - "src/pipecat/audio/**"
+      - "src/pipecat/turns/**"
+      - "src/pipecat/observers/**"
+      - "src/pipecat/pipeline/**"
+  workflow_dispatch:
+    inputs:
+      pr_number:
+        description: "PR number to generate docs for"
+        required: true
+        type: string
+
+jobs:
+  update-docs:
+    if: >-
+      github.event_name == 'workflow_dispatch' ||
+      github.event.pull_request.merged == true
+    runs-on: ubuntu-latest
+    timeout-minutes: 15
+    permissions:
+      contents: read
+      pull-requests: read
+      id-token: write
+    steps:
+      - name: Checkout pipecat
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: Checkout docs
+        uses: actions/checkout@v4
+        with:
+          repository: pipecat-ai/docs
+          token: ${{ secrets.DOCS_SYNC_TOKEN }}
+          path: _docs
+
+      - name: Resolve PR number
+        id: pr
+        run: |
+          if [ "${{ github.event_name }}" = "workflow_dispatch" ]; then
+            echo "number=${{ inputs.pr_number }}" >> "$GITHUB_OUTPUT"
+          else
+            echo "number=${{ github.event.pull_request.number }}" >> "$GITHUB_OUTPUT"
+          fi
+
+      - name: Update documentation
+        uses: anthropics/claude-code-action@v1
+        env:
+          DOCS_SYNC_TOKEN: ${{ secrets.DOCS_SYNC_TOKEN }}
+        with:
+          anthropic_api_key: ${{ secrets.ANTHROPIC_API_KEY }}
+          github_token: ${{ secrets.GITHUB_TOKEN }}
+          prompt: |
+            You are updating documentation for the pipecat-ai/docs repository based on
+            changes merged in PR #${{ steps.pr.outputs.number }} of pipecat-ai/pipecat.
+
+            ## Setup
+
+            1. Read the skill instructions at `.claude/skills/update-docs/SKILL.md`
+            2. Read the source-to-doc mapping at `.claude/skills/update-docs/SOURCE_DOC_MAPPING.md`
+            3. The docs repository is checked out at `./_docs/`
+
+            ## Get the diff
+
+            Run `gh pr diff ${{ steps.pr.outputs.number }}` to see what changed in the PR.
+            Also run `gh pr diff ${{ steps.pr.outputs.number }} --name-only` to get the list of changed files.
+            Filter to source files matching the directories listed in SKILL.md Step 3.
+
+            If no relevant source files were changed, exit with "No documentation changes needed."
+
+            ## Follow the skill instructions
+
+            Apply the SKILL.md workflow (Steps 3-9) with these adaptations for automation:
+
+            ### Docs path
+            Use `./_docs/` — it's already checked out. Do not ask for a path.
+
+            ### Branch management
+            - Branch name: `docs/pr-${{ steps.pr.outputs.number }}`
+            - Work inside `./_docs/` for all doc edits and git operations
+            - Check if the branch already exists on the remote:
+              ```bash
+              cd _docs && git fetch origin docs/pr-${{ steps.pr.outputs.number }} 2>/dev/null
+              ```
+              - If it exists: check it out (supports workflow re-runs)
+              - If not: create it from main
+
+            ### Git config
+            Before committing in `_docs`, set:
+            ```bash
+            git config user.name "github-actions[bot]"
+            git config user.email "github-actions[bot]@users.noreply.github.com"
+            ```
+
+            ### No interactive questions
+            Do not ask questions. If you encounter gaps (unmapped files, missing sections,
+            ambiguous changes), note them in the PR body under "## Gaps identified".
+
+            ### Creating the docs PR
+            After committing all changes in `_docs`, push and create a PR:
+            ```bash
+            cd _docs
+            git push -u origin docs/pr-${{ steps.pr.outputs.number }}
+            GH_TOKEN=$DOCS_SYNC_TOKEN gh pr create \
+              --repo pipecat-ai/docs \
+              --label auto-docs \
+              --label pipecat \
+              --title "docs: update for pipecat PR #${{ steps.pr.outputs.number }}" \
+              --body "$(cat <<'BODY'
+            Automated documentation update for [pipecat PR #${{ steps.pr.outputs.number }}](https://github.com/pipecat-ai/pipecat/pull/${{ steps.pr.outputs.number }}).
+
+            ## Changes
+            <summarize each doc page updated and what changed>
+
+            ## Gaps identified
+            <any unmapped files, missing doc pages, or missing sections — or "None">
+            BODY
+            )"
+            ```
+
+            ### Re-run handling
+            If `gh pr create` fails because a PR from that branch already exists,
+            push the updated commits and use `gh pr edit` to update the body instead.
+
+            ### No-op
+            If after analyzing the diff you determine no documentation changes are needed
+            (e.g., only skip-listed files changed, or changes don't affect public API docs),
+            exit cleanly without creating a branch or PR. Output "No documentation changes needed."
+
+            ## Important rules
+            - Only modify files inside `./_docs/` — never modify pipecat source code
+            - Follow the conservative editing rules from SKILL.md Step 6
+            - Read each doc page fully before editing (SKILL.md Guidelines)
+            - Use `GH_TOKEN=$DOCS_SYNC_TOKEN` for all `gh` commands targeting pipecat-ai/docs
+          claude_args: |
+            --model claude-sonnet-4-5-20250929
+            --max-turns 30
+            --allowedTools "Read,Write,Edit,Glob,Grep,Bash"

.gitignore

@@ -4,7 +4,14 @@ __pycache__/
 *~
 venv
 .venv
-/.idea
+.idea
+.gradle
+.next
+next-env.d.ts
+local.properties
+*.log
+*.lock
+smart_turn_audio_log
 #*#
 
 # Distribution / Packaging
@@ -27,7 +34,7 @@ share/python-wheels/
 *.egg
 MANIFEST
 .DS_Store
-.env
+.env*
 fly.toml
 
 # Examples
@@ -51,4 +58,7 @@ docs/api/_build/
 docs/api/api
 
 # uv
-.python-version
+.python-version
+
+# Pipecat
+whisker_setup.py

.pre-commit-config.yaml

@@ -1,8 +1,13 @@
 repos:
-  - repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: v0.12.1
+  - repo: local
     hooks:
       - id: ruff
-        language_version: python3
-        args: [--fix]
+        name: ruff
+        entry: uv run ruff check --fix
+        language: system
+        types: [python]
       - id: ruff-format
+        name: ruff-format
+        entry: uv run ruff format
+        language: system
+        types: [python]

.readthedocs.yaml

@@ -11,7 +11,7 @@ build:
   jobs:
     post_install:
       - pip install uv
-      - UV_PROJECT_ENVIRONMENT=$READTHEDOCS_VIRTUALENV_PATH uv sync --group docs --all-extras --no-extra krisp --no-extra gstreamer --no-extra ultravox --no-extra local_smart_turn --no-extra moondream --no-extra riva --no-extra mlx-whisper
+      - UV_PROJECT_ENVIRONMENT=$READTHEDOCS_VIRTUALENV_PATH uv sync --group docs --all-extras --no-extra gstreamer --no-extra local_smart_turn --no-extra moondream --no-extra mlx-whisper
 
 sphinx:
   configuration: docs/api/conf.py

AGENTS.md

@@ -0,0 +1,174 @@
+# AGENTS.md
+
+This file provides guidance to AI coding agents when working with code in this repository.
+
+## Project Overview
+
+Pipecat is an open-source Python framework for building real-time voice and multimodal conversational AI agents. It orchestrates audio/video, AI services, transports, and conversation pipelines using a frame-based architecture.
+
+## Common Commands
+
+```bash
+# Setup development environment
+uv sync --group dev --all-extras --no-extra gstreamer --no-extra local
+
+# Install pre-commit hooks
+uv run pre-commit install
+
+# Run all tests
+uv run pytest
+
+# Run a single test file
+uv run pytest tests/test_name.py
+
+# Run a specific test
+uv run pytest tests/test_name.py::test_function_name
+
+# Preview changelog
+uv run towncrier build --draft --version Unreleased
+
+# Lint and format check
+uv run ruff check
+uv run ruff format --check
+
+# Update dependencies (after editing pyproject.toml)
+uv lock && uv sync
+```
+
+## Architecture
+
+### Frame-Based Pipeline Processing
+
+All data flows as **Frame** objects through a pipeline of **FrameProcessors**:
+
+```
+[Processor1] → [Processor2] → ... → [ProcessorN]
+```
+
+**Key components:**
+
+- **Frames** (`src/pipecat/frames/frames.py`): Data units (audio, text, video) and control signals. Flow DOWNSTREAM (input→output) or UPSTREAM (acknowledgments/errors).
+
+- **FrameProcessor** (`src/pipecat/processors/frame_processor.py`): Base processing unit. Each processor receives frames, processes them, and pushes results downstream.
+
+- **Pipeline** (`src/pipecat/pipeline/pipeline.py`): Chains processors together.
+
+- **ParallelPipeline** (`src/pipecat/pipeline/parallel_pipeline.py`): Runs multiple pipelines in parallel.
+
+- **Transports** (`src/pipecat/transports/`): Transports are frame processors used for external I/O layer (Daily WebRTC, LiveKit WebRTC, WebSocket, Local). Abstract interface via `BaseTransport`, `BaseInputTransport` and `BaseOutputTransport`.
+
+- **Pipeline Task (`src/pipecat/pipeline/task.py`)**: Runs and manages a pipeline. Pipeline tasks send the first frame, `StartFrame`, to the pipeline in order for processors to know they can start processing and pushing frames. Pipeline tasks internally create a pipeline with two additional processors, a source processor before the user-defined pipeline and a sink processor at the end. Those are used for multiple things: error handling, pipeline task level events, heartbeat monitoring, etc.
+
+- **Pipeline Runner (`src/pipecat/pipeline/runner.py`)**: High-level entry point for executing pipeline tasks. Handles signal management (SIGINT/SIGTERM) for graceful shutdown and optional garbage collection. Run a single pipeline task with `await runner.run(task)` or multiple concurrently with `await asyncio.gather(runner.run(task1), runner.run(task2))`.
+
+- **Services** (`src/pipecat/services/`): 60+ AI provider integrations (STT, TTS, LLM, etc.). Extend base classes: `AIService`, `LLMService`, `STTService`, `TTSService`, `VisionService`.
+
+- **Serializers** (`src/pipecat/serializers/`): Convert frames to/from wire formats for WebSocket transports. `FrameSerializer` base class defines `serialize()` and `deserialize()`. Telephony serializers (Twilio, Plivo, Vonage, Telnyx, Exotel, Genesys) handle provider-specific protocols and audio encoding (e.g., μ-law).
+
+- **RTVI** (`src/pipecat/processors/frameworks/rtvi.py`): Real-Time Voice Interface protocol bridging clients and the pipeline. `RTVIProcessor` handles incoming client messages (text input, audio, function call results). `RTVIObserver` converts pipeline frames to outgoing messages: user/bot speaking events, transcriptions, LLM/TTS lifecycle, function calls, metrics, and audio levels.
+
+- **Observers** (`src/pipecat/observers/`): Monitor frame flow without modifying the pipeline. Passed to `PipelineTask` via the `observers` parameter. Implement `on_process_frame()` and `on_push_frame()` callbacks.
+
+### Important Patterns
+
+- **Context Aggregation**: `LLMContext` accumulates messages for LLM calls; `UserResponse` aggregates user input
+
+- **Turn Management**: Turn management is done through `LLMUserAggregator` and
+  `LLMAssistantAggregator`, created with `LLMContextAggregatorPair`
+
+- **User turn strategies**: Detection of when the user starts and stops speaking is done via user turn start/stop strategies. They push `UserStartedSpeakingFrame` and `UserStoppedSpeakingFrame` respectively.
+
+- **Interruptions**: Interruptions are usually triggered by a user turn start strategy (e.g. `VADUserTurnStartStrategy`) but they can be triggered by other processors as well, in which case the user turn start strategies don't need to. An `InterruptionFrame` carries an optional `asyncio.Event` that is set when the frame reaches the pipeline sink. If a processor stops an `InterruptionFrame` from propagating downstream (i.e., doesn't push it), it **must** call `frame.complete()` to avoid stalling `push_interruption_task_frame_and_wait()` callers.
+
+- **Uninterruptible Frames**: These are frames that will not be removed from internal queues even if there's an interruption. For example, `EndFrame` and `StopFrame`.
+
+- **Events**: Most classes in Pipecat have `BaseObject` as the very base class. `BaseObject` has support for events. Events can run in the background in an async task (default) or synchronously (`sync=True`) if we want immediate action. Synchronous event handlers need to execute fast.
+
+- **Async Task Management**: Always use `self.create_task(coroutine, name)` instead of raw `asyncio.create_task()`. The `TaskManager` automatically tracks tasks and cleans them up on processor shutdown. Use `await self.cancel_task(task, timeout)` for cancellation.
+
+- **Error Handling**: Use `await self.push_error(msg, exception, fatal)` to push errors upstream. Services should use `fatal=False` (the default) so application code can handle errors and take action (e.g. switch to another service).
+
+### Key Directories
+
+| Directory                  | Purpose                                            |
+| -------------------------- | -------------------------------------------------- |
+| `src/pipecat/frames/`      | Frame definitions (100+ types)                     |
+| `src/pipecat/processors/`  | FrameProcessor base + aggregators, filters, audio  |
+| `src/pipecat/pipeline/`    | Pipeline orchestration                             |
+| `src/pipecat/services/`    | AI service integrations (60+ providers)            |
+| `src/pipecat/transports/`  | Transport layer (Daily, LiveKit, WebSocket, Local) |
+| `src/pipecat/serializers/` | Frame serialization for WebSocket protocols        |
+| `src/pipecat/observers/`   | Pipeline observers for monitoring frame flow       |
+| `src/pipecat/audio/`       | VAD, filters, mixers, turn detection, DTMF         |
+| `src/pipecat/turns/`       | User turn management                               |
+
+## Code Style
+
+- **Docstrings**: Google-style. Classes describe purpose; `__init__` has `Args:` section; dataclasses use `Parameters:` section.
+- **Deprecations**: Use the `.. deprecated:: <version>` Sphinx directive in docstrings (never inline tags like `[DEPRECATED]`), and pair it with a runtime `warnings.warn(..., DeprecationWarning)` at the call site. See `CONTRIBUTING.md` for full conventions.
+- **Linting**: Ruff (line length 100). Pre-commit hooks enforce formatting.
+- **Type hints**: Required for complex async code.
+- **Dataclass vs Pydantic**: Use `@dataclass` for frames and internal pipeline data (high-frequency, no validation needed). Use Pydantic `BaseModel` for configuration, parameters, metrics, and external API data (benefits from validation and serialization). Specifically:
+  - `@dataclass`: Frame types, context aggregator pairs, internal data containers
+  - `BaseModel`: Service `InputParams`, transport/VAD/turn params, metrics data, API request/response models, serializer params
+
+### Docstring Example
+
+```python
+class MyService(LLMService):
+    """Description of what the service does.
+
+    More detailed description.
+
+    Event handlers available:
+
+    - on_connected: Called when we are connected
+
+    Example::
+
+        @service.event_handler("on_connected")
+        async def on_connected(service, frame):
+            ...
+    """
+
+    def __init__(self, param1: str, **kwargs):
+        """Initialize the service.
+
+        Args:
+            param1: Description of param1.
+            **kwargs: Additional arguments passed to parent.
+        """
+        super().__init__(**kwargs)
+
+
+# Pydantic params class with a deprecated field
+class MyParams(BaseModel):
+    """Configuration parameters for MyService.
+
+    Parameters:
+        new_setting: Replacement for ``old_setting``.
+        old_setting: Legacy setting, no longer used.
+
+            .. deprecated:: 1.2.0
+                Use ``new_setting`` instead. Will be removed in 2.0.0.
+    """
+
+    new_setting: str = "default"
+    old_setting: str | None = None
+```
+
+## Service Implementation
+
+When adding a new service:
+
+1. Extend the appropriate base class (`STTService`, `TTSService`, `LLMService`, etc.)
+2. Implement required abstract methods
+3. Handle necessary frames
+4. By default, all frames should be pushed in the direction they came
+5. Push `ErrorFrame` on failures
+6. Add metrics tracking via `MetricsData` if relevant
+7. Follow the pattern of existing services in `src/pipecat/services/`
+
+## Testing
+
+Test utilities live in `src/pipecat/tests/utils.py`. Use `run_test()` to send frames through a pipeline and assert expected output frames in each direction. Use `SleepFrame(sleep=N)` to add delays between frames.

CHANGELOG.md.template

@@ -1,62 +0,0 @@
-# Changelog
-
-All notable changes to the **<project name>** SDK will be documented in this file.
-
-The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
-and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
-
-Please make sure to add your changes to the appropriate categories:
-
-## [Unreleased]
-
-### Added
-
-<!-- for new functionality -->
-
-- n/a
-
-### Changed
-
-<!-- for changed functionality -->
-
-- n/a
-
-### Deprecated
-
-<!-- for soon-to-be removed functionality -->
-
-- n/a
-
-### Removed
-
-<!-- for removed functionality -->
-
-- n/a
-
-### Fixed
-
-<!-- for fixed bugs -->
-
-- n/a
-
-### Performance
-
-<!-- for performance-relevant changes -->
-
-- n/a
-
-### Security
-
-<!-- for security-relevant changes -->
-
-- n/a
-
-### Other
-
-<!-- for everything else -->
-
-- n/a
-
-## [0.1.0] - YYYY-MM-DD
-
-Initial release.

CLAUDE.md

@@ -0,0 +1 @@
+@AGENTS.md

COMMUNITY_INTEGRATIONS.md

@@ -23,9 +23,8 @@ Create your integration following the patterns and examples shown in the "Integr
 Your repository must contain these components:
 
 - **Source code** - Complete implementation following Pipecat patterns
-- **Foundational example** - Single file example showing basic usage (see [Pipecat examples](https://github.com/pipecat-ai/pipecat/tree/main/examples/foundational))
+- **Foundational example** - Single file example showing basic usage (see [Pipecat examples](https://github.com/pipecat-ai/pipecat/tree/main/examples))
 - **README.md** - Must include:
-
   - Introduction and explanation of your integration
   - Installation instructions
   - Usage instructions with Pipecat Pipeline
@@ -66,12 +65,25 @@ Once your PR is submitted, post in the `#community-integrations` Discord channel
 
 #### Websocket-based Services
 
+**Base class:** `WebsocketSTTService`
+
+**Use for:** Services where you manage the websocket connection directly. Combines `STTService` with `WebsocketService` for automatic reconnection and keepalive support.
+
+**Examples:**
+
+- [CartesiaSTTService](https://github.com/pipecat-ai/pipecat/blob/main/src/pipecat/services/cartesia/stt.py)
+- [ElevenLabsRealtimeSTTService](https://github.com/pipecat-ai/pipecat/blob/main/src/pipecat/services/elevenlabs/stt.py)
+
+#### SDK-based Streaming Services
+
 **Base class:** `STTService`
 
+**Use for:** Streaming services where the provider's Python SDK manages the connection internally.
+
 **Examples:**
 
 - [DeepgramSTTService](https://github.com/pipecat-ai/pipecat/blob/main/src/pipecat/services/deepgram/stt.py)
-- [SpeechmaticsSTTService](https://github.com/pipecat-ai/pipecat/blob/main/src/pipecat/services/speechmatics/stt.py)
+- [GoogleSTTService](https://github.com/pipecat-ai/pipecat/blob/main/src/pipecat/services/google/stt.py)
 
 #### File-based Services
 
@@ -79,7 +91,7 @@ Once your PR is submitted, post in the `#community-integrations` Discord channel
 
 **Examples:**
 
-- [RivaSTTService](https://github.com/pipecat-ai/pipecat/blob/main/src/pipecat/services/riva/stt.py)
+- [NvidiaSTTService](https://github.com/pipecat-ai/pipecat/blob/main/src/pipecat/services/nvidia/stt.py)
 - [FalSTTService](https://github.com/pipecat-ai/pipecat/blob/main/src/pipecat/services/fal/stt.py)
 
 #### Key requirements:
@@ -109,56 +121,59 @@ Once your PR is submitted, post in the `#community-integrations` Discord channel
 
 #### Key requirements:
 
+- **`_process_context(self, context: LLMContext)`** — The main method that processes an LLM context and generates a response. Each LLM service overrides `process_frame` to extract context from `LLMContextFrame` and calls `_process_context`.
+
+- **`adapter_class`** — Class attribute pointing to a `BaseLLMAdapter` subclass. Defaults to `OpenAILLMAdapter`. Non-OpenAI services must implement their own adapter (see `src/pipecat/adapters/base_llm_adapter.py`) with methods:
+  - `get_llm_invocation_params(context)` — Extract provider-specific params from universal context
+  - `to_provider_tools_format(tools_schema)` — Convert standard tools to provider format
+  - `get_messages_for_logging(context)` — Format messages for logging
+  - Reference adapters: `src/pipecat/adapters/services/` (anthropic, gemini, bedrock, etc.)
+
 - **Frame sequence:** Output must follow this frame sequence pattern:
+  - `LLMFullResponseStartFrame` — Signals the start of an LLM response
+  - `LLMTextFrame` — Contains LLM content, typically streamed as tokens
+  - `LLMFullResponseEndFrame` — Signals the end of an LLM response
 
-  - `LLMFullResponseStartFrame` - Signals the start of an LLM response
-  - `LLMTextFrame` - Contains LLM content, typically streamed as tokens
-  - `LLMFullResponseEndFrame` - Signals the end of an LLM response
+- **Thought frames (reasoning models):** If the model supports extended thinking / chain-of-thought, emit thought frames alongside the response:
+  - `LLMThoughtStartFrame` — Signals the start of a thought
+  - `LLMThoughtTextFrame` — Contains thought content, streamed as tokens
+  - `LLMThoughtEndFrame` — Signals the end of a thought
 
-- **Context aggregation:** Implement context aggregation to collect user and assistant content:
-  - Aggregators come in pairs with a `user()` instance and `assistant()` instance
-  - Context must adhere to the `LLMContext` universal format
-  - Aggregators should handle adding messages, function calls, and images to the context
+- **Context aggregation** is handled by the framework via `LLMContext` + `LLMContextAggregatorPair`. The LLM service just processes context it receives — no need to implement aggregators.
 
 ### TTS (Text-to-Speech) Services
 
-#### AudioContextWordTTSService
+#### WebsocketTTSService
 
-**Use for:** Websocket-based services supporting word/timestamp alignment
+**Use for:** Websocket-based streaming services (with or without word timestamps)
 
-**Example:**
+**Examples:**
 
 - [CartesiaTTSService](https://github.com/pipecat-ai/pipecat/blob/main/src/pipecat/services/cartesia/tts.py)
+- [ElevenLabsTTSService](https://github.com/pipecat-ai/pipecat/blob/main/src/pipecat/services/elevenlabs/tts.py)
 
 #### InterruptibleTTSService
 
-**Use for:** Websocket-based services without word/timestamp alignment, requiring disconnection on interruption
+**Use for:** Websocket-based services without word timestamps that reconnect on interruption (e.g. don't support a context ID or interruption message)
 
 **Example:**
 
 - [SarvamTTSService](https://github.com/pipecat-ai/pipecat/blob/main/src/pipecat/services/sarvam/tts.py)
 
-#### WordTTSService
-
-**Use for:** HTTP-based services supporting word/timestamp alignment
-
-**Example:**
-
-- [ElevenLabsHttpTTSService](https://github.com/pipecat-ai/pipecat/blob/main/src/pipecat/services/elevenlabs/tts.py)
-
 #### TTSService
 
-**Use for:** HTTP-based services without word/timestamp alignment
+**Use for:** HTTP-based services (word timestamps are supported in the base class)
 
-**Example:**
+**Examples:**
 
 - [GoogleHttpTTSService](https://github.com/pipecat-ai/pipecat/blob/main/src/pipecat/services/google/tts.py)
+- [OpenAITTSService](https://github.com/pipecat-ai/pipecat/blob/main/src/pipecat/services/openai/tts.py)
 
 #### Key requirements:
 
-- For websocket services, use asyncio WebSocket implementation (required for v13+ support)
+- For websocket services, use asyncio WebSocket implementation
 - Handle idle service timeouts with keepalives
-- TTSServices push both audio (`TTSRawAudioFrame`) and text (`TTSTextFrame`) frames
+- TTS services push both audio (`TTSAudioRawFrame`) and text (`TTSTextFrame`) frames
 
 ### Telephony Serializers
 
@@ -202,14 +217,25 @@ Vision services process images and provide analysis such as descriptions, object
 
 #### Key requirements:
 
-- Must implement `run_vision` method that takes an `LLMContext` and returns an `AsyncGenerator[Frame, None]`
-- The method processes the latest image in the context and yields frames with analysis results
-- Typically yields `TextFrame` objects containing descriptions or answers
+- Must implement `run_vision` method that takes a `UserImageRawFrame` and returns an `AsyncGenerator[Frame, None]`
+- The method processes the image frame and yields frames with analysis results
+- Must yield the frame sequence: `VisionFullResponseStartFrame`, `VisionTextFrame`, `VisionFullResponseEndFrame`
 
 ## Implementation Guidelines
 
 ### Naming Conventions
 
+#### Package and Repository Naming
+
+Use the `pipecat-{vendor}` naming convention for your PyPI package and repository:
+
+- `pipecat-{vendor}` — for single-service integrations (e.g., `pipecat-deepdub`)
+- `pipecat-{vendor}-{type}` — when a vendor offers multiple service types (e.g., `pipecat-upliftai-stt`, `pipecat-upliftai-tts`)
+
+This convention makes community packages easily discoverable via PyPI search and clearly identifies them as part of the Pipecat ecosystem.
+
+#### Class Naming
+
 - **STT:** `VendorSTTService`
 - **LLM:** `VendorLLMService`
 - **TTS:**
@@ -233,24 +259,137 @@ def can_generate_metrics(self) -> bool:
     return True
 ```
 
-### Dynamic Settings Updates
+### Service Settings
 
-STT, LLM, and TTS services support `ServiceUpdateSettingsFrame` for dynamic configuration changes. The base STTService has an `_update_settings()` method that handles settings, and the private `_settings` `Dict` is used to store settings and provide access to the subclass.
+Every AI service (STT, LLM, TTS, image generation, etc.) exposes a **Settings dataclass** that serves two roles:
+
+1. **Store mode** — the service's `self._settings` holds the current value of every runtime-updatable field.
+2. **Delta mode** — an update frame (e.g. `TTSUpdateSettingsFrame`) specifies only the fields that should change; unspecified fields remain `NOT_GIVEN`.
+
+#### Defining your Settings class
+
+Extend `STTSettings`, `TTSSettings`, `LLMSettings`, or `ImageGenSettings` (or, if your service directly subclasses `AIService`, `ServiceSettings`). The base classes already provide common fields (e.g. `model`, `voice`, `language`). You only need to add **service-specific knobs that should be runtime-updatable**:
 
 ```python
-async def set_language(self, language: Language):
-    """Set the recognition language and reconnect.
+from dataclasses import dataclass, field
 
-    Args:
-        language: The language to use for speech recognition.
+from pipecat.services.settings import TTSSettings, NOT_GIVEN
+
+@dataclass
+class MyTTSSettings(TTSSettings):
+    """Settings for MyTTS service.
+
+    Parameters:
+        speaking_rate: Speed multiplier (0.5–2.0).
     """
-    logger.info(f"Switching STT language to: [{language}]")
-    self._settings["language"] = language
-    await self._disconnect()
-    await self._connect()
+
+    speaking_rate: float | None = field(default_factory=lambda: NOT_GIVEN)
 ```
 
-Note that, in this example, Deepgram requires the websocket connection be disconnected and reconnected to reinitialize the service with the new value. Consider if your service requires reconnection.
+**What goes in Settings vs. `__init__` params:**
+
+| Belongs in Settings                                      | Stays as `__init__` params                |
+| -------------------------------------------------------- | ----------------------------------------- |
+| Model name, voice, language                              | API keys, auth tokens                     |
+| Service-specific tuning knobs (rate, pitch, temperature) | Base URLs, endpoint overrides             |
+| Anything users may want to change mid-session            | Audio encoding, sample format             |
+|                                                          | Connection parameters (timeouts, retries) |
+
+The rule of thumb: if a caller might send an update frame to change it at runtime, it belongs in Settings. Everything else is init-only config stored as `self._xxx`.
+
+#### Wiring settings into `__init__`
+
+Accept an **optional** `settings` parameter. Build a `default_settings` object with all fields set to real values, then merge any caller overrides with `apply_update`.
+
+Add a `Settings` **class attribute** that points to your settings dataclass. This lets callers access the settings class through the service itself (e.g. `MyTTSService.Settings(...)`) without a separate import:
+
+```python
+from typing import Optional
+
+class MyTTSService(TTSService):
+    Settings = MyTTSSettings
+    _settings: Settings
+
+    def __init__(
+        self,
+        *,
+        api_key: str,
+        settings: Optional[Settings] = None,
+        **kwargs,
+    ):
+        # 1. Defaults — every field has a real value (store mode).
+        default_settings = self.Settings(
+            model="my-model-v1",
+            voice="default-voice",
+            language="en",
+            speaking_rate=1.0,
+        )
+
+        # 2. Merge caller overrides (only given fields win).
+        if settings is not None:
+            default_settings.apply_update(settings)
+
+        # 3. Pass the fully-populated settings to the base class.
+        super().__init__(settings=default_settings, **kwargs)
+
+        # 4. Init-only config stored separately.
+        self._api_key = api_key
+```
+
+This pattern lets callers override only what they care about:
+
+```python
+# Uses all defaults
+svc = MyTTSService(api_key="sk-xxx")
+
+# Overrides just the voice — access Settings through the service class
+svc = MyTTSService(
+    api_key="sk-xxx",
+    settings=MyTTSService.Settings(voice="custom-voice"),
+)
+```
+
+#### Reacting to runtime changes
+
+AI services support runtime configuration changes via `*UpdateSettingsFrame`s (e.g. `STTUpdateSettingsFrame`, `TTSUpdateSettingsFrame`, `LLMUpdateSettingsFrame`).
+
+To react to runtime setting changes, override `_update_settings`. The base implementation applies the delta to `self._settings` and returns a `dict` mapping each changed field name to its **pre-update** value. Your override should call `super()` first, then act on the changed fields. A common implementation might look like:
+
+```python
+async def _update_settings(self, update: TTSSettings) -> dict[str, Any]:
+    """Apply a settings update, reconfiguring the connection if needed."""
+    changed = await super()._update_settings(update)
+
+    if not changed:
+        return changed
+
+    await self._disconnect()
+    await self._connect()
+
+    return changed
+```
+
+The dict keys work like a set for membership tests (`"language" in changed`) and truthiness (`if changed`). Use `changed.keys() - {"language"}` for set difference, or `changed["language"]` to inspect the previous value of a field.
+
+Note that, in this example, the service requires a reconnect to apply the new language. Consider, for each setting, whether your service requires reconnection or can apply changes in-place.
+
+If your service can't yet apply certain settings at runtime, call `self._warn_unhandled_updated_settings(changed)` with any unhandled field names so users get a clear log message:
+
+```python
+async def _update_settings(self, update: TTSSettings) -> dict[str, Any]:
+    changed = await super()._update_settings(update)
+
+    if not changed:
+        return changed
+
+    if "language" in changed:
+        await self._update_language()
+    else:
+        # TODO: this should be temporary - handle changes to other settings soon!
+        self._warn_unhandled_updated_settings(changed.keys() - {"language"})
+
+    return changed
+```
 
 ### Sample Rate Handling
 
@@ -260,7 +399,7 @@ Sample rates are set via PipelineParams and passed to each frame processor at in
 async def start(self, frame: StartFrame):
     """Start the service."""
     await super().start(frame)
-    self._settings["output_format"]["sample_rate"] = self.sample_rate
+    self._settings.output_sample_rate = self.sample_rate
     await self._connect()
 ```
 
@@ -270,7 +409,7 @@ Note that `self.sample_rate` is a `@property` set in the TTSService base class,
 
 Use Pipecat's tracing decorators:
 
-- **STT:** `@traced_stt` - decorate a function that handles `transcript`, `is_final`, `language` as args
+- **STT:** `@traced_stt` - decorate `_handle_transcription(self, transcript, is_final, language)` (the standard method name convention)
 - **LLM:** `@traced_llm` - decorate the `_process_context()` method
 - **TTS:** `@traced_tts` - decorate the `run_tts()` method
 
@@ -278,8 +417,9 @@ Use Pipecat's tracing decorators:
 
 ### Packaging and Distribution
 
+- Name your package `pipecat-{vendor}` (see [Naming Conventions](#naming-conventions))
 - Use [uv](https://docs.astral.sh/uv/) for packaging (encouraged)
-- Consider releasing to PyPI for easier installation
+- Publish to PyPI for easier installation
 - Follow semantic versioning principles
 - Maintain a changelog
 
@@ -292,17 +432,15 @@ For REST-based communication, use aiohttp. Pipecat includes this as a required d
 - Wrap API calls in appropriate try/catch blocks
 - Handle rate limits and network failures gracefully
 - Provide meaningful error messages
-- When errors occur, raise exceptions AND push `ErrorFrame`s to notify the pipeline:
+- When errors occur, raise exceptions AND push errors to notify the pipeline:
 
 ```python
-from pipecat.frames.frames import ErrorFrame
-
 try:
     # Your API call
     result = await self._make_api_call()
 except Exception as e:
-    # Push error frame to pipeline
-    await self.push_error(ErrorFrame(error=f"{self} error: {e}"))
+    # Push error upstream to notify the pipeline
+    await self.push_error(f"{self} error: {e}", exception=e)
     # Raise or handle as appropriate
     raise
 ```

CONTRIBUTING.md

@@ -17,24 +17,120 @@ We welcome contributions of all kinds! Your help is appreciated. Follow these st
    git checkout -b your-branch-name
    ```
 4. **Make your changes**: Edit or add files as necessary.
-5. **Test your changes**: Ensure that your changes look correct and follow the style set in the codebase.
-6. **Commit your changes**: Once you're satisfied with your changes, commit them with a meaningful message.
+5. **Add a changelog entry**: Create a changelog fragment file (see [Changelog Entries](#changelog-entries) below).
+6. **Test your changes**: Ensure that your changes look correct and follow the style set in the codebase.
+7. **Commit your changes**: Once you're satisfied with your changes, commit them with a meaningful message.
 
 ```bash
 git commit -m "Description of your changes"
 ```
 
-7. **Push your changes**: Push your branch to your forked repository.
+8. **Push your changes**: Push your branch to your forked repository.
 
 ```bash
 git push origin your-branch-name
 ```
 
-8. **Submit a Pull Request (PR)**: Open a PR from your forked repository to the main branch of this repo.
+9. **Submit a Pull Request (PR)**: Open a PR from your forked repository to the main branch of this repo.
    > Important: Describe the changes you've made clearly!
 
 Our maintainers will review your PR, and once everything is good, your contributions will be merged!
 
+## Changelog Entries
+
+Every pull request that makes a user-facing change should include a changelog entry. We use a changelog fragment system to avoid merge conflicts.
+
+### Creating a Changelog Fragment
+
+1. Create a new file in the `changelog/` directory with this naming pattern:
+
+   ```
+   <PR_number>.<type>.md
+   ```
+
+2. Choose the appropriate type:
+   - `added.md` - New features
+   - `changed.md` - Changes in existing functionality
+   - `deprecated.md` - Soon-to-be removed features
+   - `removed.md` - Removed features
+   - `fixed.md` - Bug fixes
+   - `performance.md` - Performance improvements
+   - `security.md` - Security fixes
+   - `other.md` - Other changes (documentation, dependencies, etc.)
+
+3. Write your changelog entry as a Markdown bullet point. Include the `-` at the start:
+
+**Example files:**
+
+`changelog/1234.added.md`:
+
+```markdown
+- Added support for Anthropic Claude 3.5 Sonnet with improved streaming performance.
+```
+
+`changelog/5678.fixed.md`:
+
+```markdown
+- Fixed an issue where audio frames were dropped during high-load scenarios.
+```
+
+**For entries with nested bullets:**
+
+`changelog/1234.changed.md`:
+
+```markdown
+- Updated service configuration:
+  - Changed default timeout to 30 seconds
+  - Added retry logic for failed connections
+```
+
+### Multiple Changes in One PR
+
+**Different types of changes:** Create separate fragment files for each type:
+
+```
+changelog/1234.added.md
+changelog/1234.fixed.md
+```
+
+**Multiple changes of the same type:** Create numbered fragment files:
+
+```
+changelog/1234.changed.md
+changelog/1234.changed.2.md
+```
+
+**Related changes:** Use nested bullets in a single fragment:
+
+```markdown
+- Updated service configuration:
+  - Changed default timeout to 30 seconds
+  - Added retry logic for failed connections
+```
+
+**Rule of thumb:** One logical change per fragment file. If changes are unrelated, use separate files.
+
+### Preview Your Changes
+
+To see what your changelog entry will look like:
+
+```bash
+towncrier build --draft --version Unreleased
+```
+
+This won't modify any files, just show you a preview.
+
+### When to Skip Changelog Entries
+
+You can skip adding a changelog entry for:
+
+- Documentation-only changes
+- Internal refactoring with no user-facing impact
+- Test-only changes
+- CI/build configuration changes
+
+If you're unsure whether your change needs a changelog entry, ask in your PR!
+
 ## Dependency Management
 
 This project uses [uv](https://docs.astral.sh/uv/) for dependency management. The `uv.lock` file is committed to ensure reproducible builds.

LICENSE

@@ -1,6 +1,6 @@
 BSD 2-Clause License
 
-Copyright (c) 2024–2025, Daily
+Copyright (c) 2024–2026, Daily
 
 Redistribution and use in source and binary forms, with or without
 modification, are permitted provided that the following conditions are met:

README.md

@@ -3,13 +3,12 @@
 </div></h1>
 
 [![PyPI](https://img.shields.io/pypi/v/pipecat-ai)](https://pypi.org/project/pipecat-ai) ![Tests](https://github.com/pipecat-ai/pipecat/actions/workflows/tests.yaml/badge.svg) [![codecov](https://codecov.io/gh/pipecat-ai/pipecat/graph/badge.svg?token=LNVUIVO4Y9)](https://codecov.io/gh/pipecat-ai/pipecat) [![Docs](https://img.shields.io/badge/Documentation-blue)](https://docs.pipecat.ai) [![Discord](https://img.shields.io/discord/1239284677165056021)](https://discord.gg/pipecat) [![Ask DeepWiki](https://deepwiki.com/badge.svg)](https://deepwiki.com/pipecat-ai/pipecat)
-[![](https://getmanta.ai/api/badges?text=Manta%20Graph&link=manta)](https://getmanta.ai/pipecat)
 
 # 🎙️ Pipecat: Real-Time Voice & Multimodal AI Agents
 
 **Pipecat** is an open-source Python framework for building real-time voice and multimodal conversational agents. Orchestrate audio and video, AI services, different transports, and conversation pipelines effortlessly—so you can focus on what makes your agent unique.
 
-> Want to dive right in? Try the [quickstart](https://docs.pipecat.ai/getting-started/quickstart).
+> Want to dive right in? Run `pipecat init quickstart` or follow the [quickstart guide](https://docs.pipecat.ai/getting-started/quickstart).
 
 ## 🚀 What You Can Build
 
@@ -29,6 +28,10 @@
 
 ## 🌐 Pipecat Ecosystem
 
+### 🧩 Multi-agent systems
+
+Need multiple AI agents working together? [Pipecat Subagents](https://github.com/pipecat-ai/pipecat-subagents) lets you build distributed multi-agent systems where each agent runs its own pipeline and communicates through a shared message bus. Hand off conversations between specialists, dispatch background tasks, and scale agents across processes or machines.
+
 ### 📱 Client SDKs
 
 Building client applications? You can connect to Pipecat from any platform using our official SDKs:
@@ -56,6 +59,20 @@ Looking for help debugging your pipeline and processors? Check out [Whisker](htt
 
 Love terminal applications? Check out [Tail](https://github.com/pipecat-ai/tail), a terminal dashboard for Pipecat.
 
+### 🤖 Claude Code Skills
+
+Use [Pipecat Skills](https://github.com/pipecat-ai/skills) with [Claude Code](https://claude.ai/code) to scaffold projects, deploy to Pipecat Cloud, and more. Install the marketplace with:
+
+```
+claude plugin marketplace add pipecat-ai/skills
+```
+
+and install any of the available plugins.
+
+### 🧩 Community Integrations
+
+Build and share your own Pipecat service integrations! Browse existing [community integrations](https://docs.pipecat.ai/api-reference/server/services/community-integrations) or check out our [guide](COMMUNITY_INTEGRATIONS.md) to create your own.
+
 ### 📺️ Pipecat TV Channel
 
 Catch new features, interviews, and how-tos on our [Pipecat TV](https://www.youtube.com/playlist?list=PLzU2zoMTQIHjqC3v4q2XVSR3hGSzwKFwH) channel.
@@ -66,27 +83,28 @@ Catch new features, interviews, and how-tos on our [Pipecat TV](https://www.yout
     <a href="https://github.com/pipecat-ai/pipecat-examples/tree/main/simple-chatbot"><img src="https://raw.githubusercontent.com/pipecat-ai/pipecat-examples/main/simple-chatbot/image.png" width="400" /></a>&nbsp;
     <a href="https://github.com/pipecat-ai/pipecat-examples/tree/main/storytelling-chatbot"><img src="https://raw.githubusercontent.com/pipecat-ai/pipecat-examples/main/storytelling-chatbot/image.png" width="400" /></a>
     <br/>
-    <a href="https://github.com/pipecat-ai/pipecat-examples/tree/main/translation-chatbot"><img src="https://raw.githubusercontent.com/pipecat-ai/pipecat-examples/main/translation-chatbot/image.png" width="400" /></a>&nbsp;
-    <a href="https://github.com/pipecat-ai/pipecat/blob/main/examples/foundational/12-describe-video.py"><img src="https://github.com/pipecat-ai/pipecat/blob/main/examples/foundational/assets/moondream.png" width="400" /></a>
+    <a href="https://github.com/pipecat-ai/pipecat-examples/tree/main/daily-multi-translation"><img src="https://raw.githubusercontent.com/pipecat-ai/pipecat-examples/main/daily-multi-translation/image.png" width="400" /></a>&nbsp;
+    <a href="https://github.com/pipecat-ai/pipecat/blob/main/examples/vision/vision-moondream.py"><img src="https://github.com/pipecat-ai/pipecat/blob/main/examples/assets/moondream.png" width="400" /></a>
 </p>
 
 ## 🧩 Available services
 
-| Category            | Services                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                  |
-| ------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
-| Speech-to-Text      | [AssemblyAI](https://docs.pipecat.ai/server/services/stt/assemblyai), [AWS](https://docs.pipecat.ai/server/services/stt/aws), [Azure](https://docs.pipecat.ai/server/services/stt/azure), [Cartesia](https://docs.pipecat.ai/server/services/stt/cartesia), [Deepgram](https://docs.pipecat.ai/server/services/stt/deepgram), [ElevenLabs](https://docs.pipecat.ai/server/services/stt/elevenlabs), [Fal Wizper](https://docs.pipecat.ai/server/services/stt/fal), [Gladia](https://docs.pipecat.ai/server/services/stt/gladia), [Google](https://docs.pipecat.ai/server/services/stt/google), [Groq (Whisper)](https://docs.pipecat.ai/server/services/stt/groq), [NVIDIA Riva](https://docs.pipecat.ai/server/services/stt/riva), [OpenAI (Whisper)](https://docs.pipecat.ai/server/services/stt/openai), [SambaNova (Whisper)](https://docs.pipecat.ai/server/services/stt/sambanova), [Sarvam](https://docs.pipecat.ai/server/services/stt/sarvam), [Soniox](https://docs.pipecat.ai/server/services/stt/soniox), [Speechmatics](https://docs.pipecat.ai/server/services/stt/speechmatics), [Ultravox](https://docs.pipecat.ai/server/services/stt/ultravox), [Whisper](https://docs.pipecat.ai/server/services/stt/whisper)                                                                                                                                                                                          |
-| LLMs                | [Anthropic](https://docs.pipecat.ai/server/services/llm/anthropic), [AWS](https://docs.pipecat.ai/server/services/llm/aws), [Azure](https://docs.pipecat.ai/server/services/llm/azure), [Cerebras](https://docs.pipecat.ai/server/services/llm/cerebras), [DeepSeek](https://docs.pipecat.ai/server/services/llm/deepseek), [Fireworks AI](https://docs.pipecat.ai/server/services/llm/fireworks), [Gemini](https://docs.pipecat.ai/server/services/llm/gemini), [Grok](https://docs.pipecat.ai/server/services/llm/grok), [Groq](https://docs.pipecat.ai/server/services/llm/groq), [Mistral](https://docs.pipecat.ai/server/services/llm/mistral), [NVIDIA NIM](https://docs.pipecat.ai/server/services/llm/nim), [Ollama](https://docs.pipecat.ai/server/services/llm/ollama), [OpenAI](https://docs.pipecat.ai/server/services/llm/openai), [OpenRouter](https://docs.pipecat.ai/server/services/llm/openrouter), [Perplexity](https://docs.pipecat.ai/server/services/llm/perplexity), [Qwen](https://docs.pipecat.ai/server/services/llm/qwen), [SambaNova](https://docs.pipecat.ai/server/services/llm/sambanova) [Together AI](https://docs.pipecat.ai/server/services/llm/together)                                                                                                                                                                                                                              |
-| Text-to-Speech      | [Async](https://docs.pipecat.ai/server/services/tts/asyncai), [AWS](https://docs.pipecat.ai/server/services/tts/aws), [Azure](https://docs.pipecat.ai/server/services/tts/azure), [Cartesia](https://docs.pipecat.ai/server/services/tts/cartesia), [Deepgram](https://docs.pipecat.ai/server/services/tts/deepgram), [ElevenLabs](https://docs.pipecat.ai/server/services/tts/elevenlabs), [Fish](https://docs.pipecat.ai/server/services/tts/fish), [Google](https://docs.pipecat.ai/server/services/tts/google), [Groq](https://docs.pipecat.ai/server/services/tts/groq), [Hume](https://docs.pipecat.ai/server/services/tts/hume), [Inworld](https://docs.pipecat.ai/server/services/tts/inworld), [LMNT](https://docs.pipecat.ai/server/services/tts/lmnt), [MiniMax](https://docs.pipecat.ai/server/services/tts/minimax), [Neuphonic](https://docs.pipecat.ai/server/services/tts/neuphonic), [NVIDIA Riva](https://docs.pipecat.ai/server/services/tts/riva), [OpenAI](https://docs.pipecat.ai/server/services/tts/openai), [Piper](https://docs.pipecat.ai/server/services/tts/piper), [PlayHT](https://docs.pipecat.ai/server/services/tts/playht), [Rime](https://docs.pipecat.ai/server/services/tts/rime), [Sarvam](https://docs.pipecat.ai/server/services/tts/sarvam), [Speechmatics](https://docs.pipecat.ai/server/services/tts/speechmatics), [XTTS](https://docs.pipecat.ai/server/services/tts/xtts) |
-| Speech-to-Speech    | [AWS Nova Sonic](https://docs.pipecat.ai/server/services/s2s/aws), [Gemini Multimodal Live](https://docs.pipecat.ai/server/services/s2s/gemini), [OpenAI Realtime](https://docs.pipecat.ai/server/services/s2s/openai)                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                    |
-| Transport           | [Daily (WebRTC)](https://docs.pipecat.ai/server/services/transport/daily), [FastAPI Websocket](https://docs.pipecat.ai/server/services/transport/fastapi-websocket), [SmallWebRTCTransport](https://docs.pipecat.ai/server/services/transport/small-webrtc), [WebSocket Server](https://docs.pipecat.ai/server/services/transport/websocket-server), Local                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                |
-| Serializers         | [Plivo](https://docs.pipecat.ai/server/utilities/serializers/plivo), [Twilio](https://docs.pipecat.ai/server/utilities/serializers/twilio), [Telnyx](https://docs.pipecat.ai/server/utilities/serializers/telnyx)                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                         |
-| Video               | [HeyGen](https://docs.pipecat.ai/server/services/video/heygen), [Tavus](https://docs.pipecat.ai/server/services/video/tavus), [Simli](https://docs.pipecat.ai/server/services/video/simli)                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                |
-| Memory              | [mem0](https://docs.pipecat.ai/server/services/memory/mem0)                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               |
-| Vision & Image      | [fal](https://docs.pipecat.ai/server/services/image-generation/fal), [Google Imagen](https://docs.pipecat.ai/server/services/image-generation/fal), [Moondream](https://docs.pipecat.ai/server/services/vision/moondream)                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                 |
-| Audio Processing    | [Silero VAD](https://docs.pipecat.ai/server/utilities/audio/silero-vad-analyzer), [Krisp](https://docs.pipecat.ai/server/utilities/audio/krisp-filter), [Koala](https://docs.pipecat.ai/server/utilities/audio/koala-filter), [ai-coustics](https://docs.pipecat.ai/server/utilities/audio/aic-filter)                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                    |
-| Analytics & Metrics | [OpenTelemetry](https://docs.pipecat.ai/server/utilities/opentelemetry), [Sentry](https://docs.pipecat.ai/server/services/analytics/sentry)                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               |
+| Category            | Services                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                            |
+| ------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| Speech-to-Text      | [AssemblyAI](https://docs.pipecat.ai/api-reference/server/services/stt/assemblyai), [AWS](https://docs.pipecat.ai/api-reference/server/services/stt/aws), [Azure](https://docs.pipecat.ai/api-reference/server/services/stt/azure), [Cartesia](https://docs.pipecat.ai/api-reference/server/services/stt/cartesia), [Deepgram](https://docs.pipecat.ai/api-reference/server/services/stt/deepgram), [ElevenLabs](https://docs.pipecat.ai/api-reference/server/services/stt/elevenlabs), [Fal Wizper](https://docs.pipecat.ai/api-reference/server/services/stt/fal), [Gladia](https://docs.pipecat.ai/api-reference/server/services/stt/gladia), [Google](https://docs.pipecat.ai/api-reference/server/services/stt/google), [Gradium](https://docs.pipecat.ai/api-reference/server/services/stt/gradium), [Groq (Whisper)](https://docs.pipecat.ai/api-reference/server/services/stt/groq), [Mistral](https://docs.pipecat.ai/api-reference/server/services/stt/mistral), [NVIDIA](https://docs.pipecat.ai/api-reference/server/services/stt/nvidia), [OpenAI (Whisper)](https://docs.pipecat.ai/api-reference/server/services/stt/openai), [Sarvam](https://docs.pipecat.ai/api-reference/server/services/stt/sarvam), [Soniox](https://docs.pipecat.ai/api-reference/server/services/stt/soniox), [Speechmatics](https://docs.pipecat.ai/api-reference/server/services/stt/speechmatics), [Whisper](https://docs.pipecat.ai/api-reference/server/services/stt/whisper), [xAI](https://docs.pipecat.ai/api-reference/server/services/stt/xai)                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                     |
+| LLMs                | [Anthropic](https://docs.pipecat.ai/api-reference/server/services/llm/anthropic), [AWS](https://docs.pipecat.ai/api-reference/server/services/llm/aws), [Azure](https://docs.pipecat.ai/api-reference/server/services/llm/azure), [Cerebras](https://docs.pipecat.ai/api-reference/server/services/llm/cerebras), [DeepSeek](https://docs.pipecat.ai/api-reference/server/services/llm/deepseek), [Fireworks AI](https://docs.pipecat.ai/api-reference/server/services/llm/fireworks), [Gemini](https://docs.pipecat.ai/api-reference/server/services/llm/gemini), [Grok](https://docs.pipecat.ai/api-reference/server/services/llm/grok), [Groq](https://docs.pipecat.ai/api-reference/server/services/llm/groq), [Inception](https://docs.pipecat.ai/api-reference/server/services/llm/inception), [Mistral](https://docs.pipecat.ai/api-reference/server/services/llm/mistral), [Nebius](https://docs.pipecat.ai/api-reference/server/services/llm/nebius), [Novita](https://docs.pipecat.ai/api-reference/server/services/llm/novita), [NVIDIA NIM](https://docs.pipecat.ai/api-reference/server/services/llm/nvidia), [Ollama](https://docs.pipecat.ai/api-reference/server/services/llm/ollama), [OpenAI](https://docs.pipecat.ai/api-reference/server/services/llm/openai), [OpenAI Responses](https://docs.pipecat.ai/api-reference/server/services/llm/openai-responses), [OpenRouter](https://docs.pipecat.ai/api-reference/server/services/llm/openrouter), [Perplexity](https://docs.pipecat.ai/api-reference/server/services/llm/perplexity), [Qwen](https://docs.pipecat.ai/api-reference/server/services/llm/qwen), [SambaNova](https://docs.pipecat.ai/api-reference/server/services/llm/sambanova), [Sarvam](https://docs.pipecat.ai/api-reference/server/services/llm/sarvam), [Together AI](https://docs.pipecat.ai/api-reference/server/services/llm/together)                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                    |
+| Text-to-Speech      | [Async](https://docs.pipecat.ai/api-reference/server/services/tts/asyncai), [AWS](https://docs.pipecat.ai/api-reference/server/services/tts/aws), [Azure](https://docs.pipecat.ai/api-reference/server/services/tts/azure), [Camb AI](https://docs.pipecat.ai/api-reference/server/services/tts/camb), [Cartesia](https://docs.pipecat.ai/api-reference/server/services/tts/cartesia), [Deepgram](https://docs.pipecat.ai/api-reference/server/services/tts/deepgram), [ElevenLabs](https://docs.pipecat.ai/api-reference/server/services/tts/elevenlabs), [Fish](https://docs.pipecat.ai/api-reference/server/services/tts/fish), [Google](https://docs.pipecat.ai/api-reference/server/services/tts/google), [Gradium](https://docs.pipecat.ai/api-reference/server/services/tts/gradium), [Groq](https://docs.pipecat.ai/api-reference/server/services/tts/groq), [Hume](https://docs.pipecat.ai/api-reference/server/services/tts/hume), [Inworld](https://docs.pipecat.ai/api-reference/server/services/tts/inworld), [Kokoro](https://docs.pipecat.ai/api-reference/server/services/tts/kokoro), [LMNT](https://docs.pipecat.ai/api-reference/server/services/tts/lmnt), [MiniMax](https://docs.pipecat.ai/api-reference/server/services/tts/minimax), [Mistral](https://docs.pipecat.ai/api-reference/server/services/tts/mistral), [Neuphonic](https://docs.pipecat.ai/api-reference/server/services/tts/neuphonic), [NVIDIA](https://docs.pipecat.ai/api-reference/server/services/tts/nvidia), [OpenAI](https://docs.pipecat.ai/api-reference/server/services/tts/openai), [Piper](https://docs.pipecat.ai/api-reference/server/services/tts/piper), [Resemble](https://docs.pipecat.ai/api-reference/server/services/tts/resemble), [Rime](https://docs.pipecat.ai/api-reference/server/services/tts/rime), [Sarvam](https://docs.pipecat.ai/api-reference/server/services/tts/sarvam), [Smallest](https://docs.pipecat.ai/api-reference/server/services/tts/smallest), [Soniox](https://docs.pipecat.ai/api-reference/server/services/tts/soniox), [Speechmatics](https://docs.pipecat.ai/api-reference/server/services/tts/speechmatics), [xAI](https://docs.pipecat.ai/api-reference/server/services/tts/xai), [XTTS](https://docs.pipecat.ai/api-reference/server/services/tts/xtts) |
+| Speech-to-Speech    | [AWS Nova Sonic](https://docs.pipecat.ai/api-reference/server/services/s2s/aws), [Gemini Multimodal Live](https://docs.pipecat.ai/api-reference/server/services/s2s/gemini), [Grok Voice Agent](https://docs.pipecat.ai/api-reference/server/services/s2s/grok), [OpenAI Realtime](https://docs.pipecat.ai/api-reference/server/services/s2s/openai), [Ultravox](https://docs.pipecat.ai/api-reference/server/services/s2s/ultravox),                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               |
+| Transport           | [Daily (WebRTC)](https://docs.pipecat.ai/api-reference/server/services/transport/daily), [FastAPI Websocket](https://docs.pipecat.ai/api-reference/server/services/transport/fastapi-websocket), [LiveKit (WebRTC)](https://docs.pipecat.ai/api-reference/server/services/transport/livekit), [SmallWebRTCTransport](https://docs.pipecat.ai/api-reference/server/services/transport/small-webrtc), [Vonage (WebRTC)](https://docs.pipecat.ai/api-reference/server/services/transport/vonage), [WebSocket Server](https://docs.pipecat.ai/api-reference/server/services/transport/websocket-server), [WhatsApp](https://docs.pipecat.ai/api-reference/server/services/transport/whatsapp), Local                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               |
+| Serializers         | [Exotel](https://docs.pipecat.ai/api-reference/server/services/serializers/exotel), [Genesys](https://docs.pipecat.ai/api-reference/server/services/serializers/genesys), [Plivo](https://docs.pipecat.ai/api-reference/server/services/serializers/plivo), [Twilio](https://docs.pipecat.ai/api-reference/server/services/serializers/twilio), [Telnyx](https://docs.pipecat.ai/api-reference/server/services/serializers/telnyx), [Vonage](https://docs.pipecat.ai/api-reference/server/services/serializers/vonage)                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                              |
+| Video               | [HeyGen](https://docs.pipecat.ai/api-reference/server/services/video/heygen), [LemonSlice](https://docs.pipecat.ai/api-reference/server/services/transport/lemonslice), [Tavus](https://docs.pipecat.ai/api-reference/server/services/video/tavus), [Simli](https://docs.pipecat.ai/api-reference/server/services/video/simli)                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                      |
+| Memory              | [mem0](https://docs.pipecat.ai/api-reference/server/services/memory/mem0)                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                           |
+| Vision & Image      | [fal](https://docs.pipecat.ai/api-reference/server/services/image-generation/fal), [Google Imagen](https://docs.pipecat.ai/api-reference/server/services/image-generation/google-imagen), [Moondream](https://docs.pipecat.ai/api-reference/server/services/vision/moondream)                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                       |
+| Audio Processing    | [Silero VAD](https://docs.pipecat.ai/api-reference/server/utilities/audio/silero-vad-analyzer), [Krisp Viva](https://docs.pipecat.ai/guides/features/krisp-viva), [Koala](https://docs.pipecat.ai/api-reference/server/utilities/audio/koala-filter), [ai-coustics](https://docs.pipecat.ai/api-reference/server/utilities/audio/aic-filter), [RNNoise](https://docs.pipecat.ai/api-reference/server/utilities/audio/rnnoise-filter)                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                |
+| Analytics & Metrics | [OpenTelemetry](https://docs.pipecat.ai/api-reference/server/utilities/opentelemetry), [Sentry](https://docs.pipecat.ai/api-reference/server/services/analytics/sentry)                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                             |
+| Community           | [Browse community integrations →](https://docs.pipecat.ai/api-reference/server/services/community-integrations)                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                     |
 
-📚 [View full services documentation →](https://docs.pipecat.ai/server/services/supported-services)
+📚 [View full services documentation →](https://docs.pipecat.ai/api-reference/server/services/supported-services)
 
 ## ⚡ Getting started
 
@@ -128,15 +146,15 @@ You can get started with Pipecat running on your local machine, then move your a
 
 ## 🧪 Code examples
 
-- [Foundational](https://github.com/pipecat-ai/pipecat/tree/main/examples/foundational) — small snippets that build on each other, introducing one or two concepts at a time
+- [Foundational](https://github.com/pipecat-ai/pipecat/tree/main/examples) — small snippets that build on each other, introducing one or two concepts at a time
 - [Example apps](https://github.com/pipecat-ai/pipecat-examples) — complete applications that you can use as starting points for development
 
 ## 🛠️ Contributing to the framework
 
 ### Prerequisites
 
-**Minimum Python Version:** 3.10
-**Recommended Python Version:** 3.12
+**Minimum Python Version:** 3.11
+**Recommended Python Version:** >= 3.12
 
 ### Setup Steps
 
@@ -152,9 +170,7 @@ You can get started with Pipecat running on your local machine, then move your a
    ```bash
    uv sync --group dev --all-extras \
      --no-extra gstreamer \
-     --no-extra krisp \
      --no-extra local \
-     --no-extra ultravox # (ultravox not fully supported on macOS)
    ```
 
 3. Install the git pre-commit hooks:
@@ -165,6 +181,15 @@ You can get started with Pipecat running on your local machine, then move your a
 
 > **Note**: Some extras (local, gstreamer) require system dependencies. See documentation if you encounter build errors.
 
+### Claude Code Skills
+
+Install development workflow skills for contributing to Pipecat with [Claude Code](https://claude.ai/code):
+
+```
+claude plugin marketplace add pipecat-ai/pipecat
+claude plugin install pipecat-dev@pipecat-dev-skills
+```
+
 ### Running tests
 
 To run all tests, from the root directory:

changelog/4052.added.md

@@ -0,0 +1 @@
+- Added `VonageVideoConnectorTransport`, a new transport integration for real-time Vonage WebRTC sessions using the Vonage Video Connector library.

changelog/4306.fixed.md

@@ -0,0 +1 @@
+- Fixed Azure TTS last word being missed by observers and RTVI UI. The completion signal was racing with word timestamp processing, causing the final word's `TTSTextFrame` to arrive after `TTSStoppedFrame`. Completion is now routed through the word boundary queue to ensure all words are processed before signaling stream end.

changelog/4380.fixed.2.md

@@ -0,0 +1 @@
+- Fixed `BaseOutputTransport` reordering frames that share the same presentation timestamp. Frames with equal PTS values are now emitted in insertion order, preventing subtle audio/text sequencing bugs when multiple frames arrive at the same time.

changelog/4380.fixed.3.md

@@ -0,0 +1 @@
+- Fixed Cartesia word timestamps leaking SSML tag text (e.g. `<spell>`, `<emotion>`, `<break>`) into word entries. Tags are now stripped before processing, so word-to-text attribution remains accurate when SSML markup is present in the TTS input.

changelog/4380.fixed.4.md

@@ -0,0 +1 @@
+- Fixed `TTSTextFrame` entries losing their original text structure when word timestamps are enabled. Each `TTSTextFrame` now carries a `raw_text` field containing the corresponding span of the original LLM-produced text (including pattern delimiters such as `<card>4111 1111 1111 1111</card>`), so the assistant context receives properly-tagged content rather than the cleaned words returned by the TTS provider. Also handles words that straddle two sentence boundaries by splitting them and attributing each part to its correct source frame.

changelog/4380.fixed.md

@@ -0,0 +1 @@
+- Fixed skipped TTS frames (e.g. code blocks filtered via `skip_aggregator_types`) being emitted to the assistant context immediately instead of waiting for preceding spoken frames to finish. They now hold their position in the frame sequence and are flushed only after all earlier spoken sentences are complete, keeping context ordering correct.

changelog/4423.added.md

@@ -0,0 +1 @@
+- Added `InceptionLLMService` for Inception's Mercury 2 diffusion reasoning model, with support for `reasoning_effort` and `realtime` settings.

changelog/4442.added.2.md

@@ -0,0 +1 @@
+- Added `GET /status` endpoint to the development runner that reports which transports the running instance accepts (all by default, or the single transport passed via `-t`).

changelog/4442.added.md

@@ -0,0 +1 @@
+- Added plain WebSocket transport support to the development runner. Bots can now accept connections from non-telephony WebSocket clients (e.g., browser apps using protobuf framing) via the `/ws-client` endpoint alongside other transports.

changelog/4442.changed.md

@@ -0,0 +1 @@
+- ⚠️ The development runner now supports all transports (WebRTC, Daily, telephony, plain WebSocket) simultaneously from a single server. The `/start` endpoint accepts a `"transport"` field to select the transport per-request; omitting `-t` at startup enables all transports instead of defaulting to WebRTC. The Daily browser-redirect route moved from `GET /` to `GET /daily`.

changelog/4507.fixed.md

@@ -0,0 +1 @@
+- Fixed `ElevenLabsSTTService` crashing when `language` was passed as `None`. When `language` is not set, the service now lets ElevenLabs auto-detect the audio language.

changelog/4514.fixed.md

@@ -0,0 +1 @@
+- Fixed websocket STT connection setup failures so services clear stale websocket state and emit non-fatal error frames, allowing `ServiceSwitcher` failover to keep agents running.

changelog/4521.added.md

@@ -0,0 +1 @@
+- Added `max_endpoint_delay_ms` to `SonioxSTTService.Settings`, controlling the maximum delay (500-3000 ms) before endpoint detection finalizes a turn.

changelog/4521.changed.md

@@ -0,0 +1 @@
+- `SonioxSTTService` now applies settings updates (e.g. via `STTUpdateSettingsFrame`) using a graceful reconnect instead of a hard disconnect/reconnect, preserving the service's reconnect retry behavior.

changelog/4521.removed.md

@@ -0,0 +1 @@
+- Removed the unsupported Georgian (`Language.KA`) language mapping from `SonioxSTTService`.

changelog/4522.changed.md

@@ -0,0 +1 @@
+- Updated the default p99 TTFS latency values for Smallest AI, Mistral, and XAI STT so turn stop timing uses measured values instead of the conservative fallback.

changelog/4524.changed.md

@@ -0,0 +1 @@
+- Updated the development runner startup banner to show the prebuilt client URL once and list enabled or disabled transports with install hints.

changelog/4524.fixed.md

@@ -0,0 +1 @@
+- Fixed the development runner so missing optional transport dependencies disable only their related routes instead of failing startup in all-transport mode.

changelog/4527.fixed.md

@@ -0,0 +1 @@
+- Fixed a race in `ElevenLabsTTSService` where the periodic keepalive could be sent for a new turn's context before that context's `voice_settings` initialization message, causing ElevenLabs to close the WebSocket with a 1008 policy violation (`voice_settings field must be provided in the first message ...`). The keepalive now only targets a context once its context-init has been sent.

changelog/4531.changed.md

@@ -0,0 +1 @@
+- Bumped `pipecat-ai-prebuilt` to 1.0.1 in the `runner` extra, updating the prebuilt client UI served by the development runner.

changelog/_template.md.j2

@@ -0,0 +1,16 @@
+{% for section, _ in sections.items() %}
+{% if sections[section] %}
+{% for category, val in definitions.items() if category in sections[section]%}
+### {{ definitions[category]['name'] }}
+
+{% for text, values in sections[section][category].items() %}
+{{ text }}
+  (PR {{ values|join(', ') }})
+
+{% endfor %}
+{% endfor %}
+{% else %}
+No significant changes.
+
+{% endif %}
+{% endfor %}

docs/api/README.md

@@ -1,109 +1,60 @@
-# Pipecat Documentation
+# Pipecat API Documentation
 
-This directory contains the source files for auto-generating Pipecat's server API reference documentation.
-
-## Setup
-
-1. Install documentation dependencies:
-
-```bash
-pip install -r requirements.txt
-```
-
-2. Make the build scripts executable:
-
-```bash
-chmod +x build-docs.sh rtd-test.py
-```
+This directory contains the source files for auto-generating Pipecat's API reference documentation.
 
 ## Building Documentation
 
-From this directory, you can build the documentation in several ways:
-
-### Local Build
+From this directory:
 
 ```bash
-# Using the build script (automatically opens docs when done)
-./build-docs.sh
+# Build docs (warnings shown but don't fail the build)
+cd docs/api && uv run ./build-docs.sh
 
-# Or directly with sphinx-build
-sphinx-build -b html . _build/html -W --keep-going
+# Build with strict mode (warnings treated as errors)
+cd docs/api && uv run ./build-docs.sh --strict
 ```
 
-### ReadTheDocs Test Build
+The build script will:
 
-To test the documentation build process exactly as it would run on ReadTheDocs:
-
-```bash
-./rtd-test.py
-```
-
-This script:
-
-- Creates a fresh virtual environment
-- Installs all dependencies as specified in requirements files
-- Handles conflicting dependencies (like grpcio versions for Riva and PlayHT)
-- Builds the documentation in an isolated environment
-- Provides detailed logging of the build process
-
-Use this script to verify your documentation will build correctly on ReadTheDocs before pushing changes.
-
-## Viewing Documentation
-
-The built documentation will be available at `_build/html/index.html`. To open:
-
-```bash
-# On MacOS
-open _build/html/index.html
-
-# On Linux
-xdg-open _build/html/index.html
-
-# On Windows
-start _build/html/index.html
-```
+1. Install documentation dependencies via `uv sync --group docs`
+2. Clean previous build output
+3. Run `sphinx-build` to generate HTML documentation
+4. Open the result in your browser (macOS)
 
 ## Directory Structure
 
 ```
 .
-├── api/            # Auto-generated API documentation
-├── _build/         # Built documentation
-├── _static/        # Static files (images, css, etc.)
-├── conf.py         # Sphinx configuration
+├── api/            # Auto-generated API documentation (created during build)
+├── _build/         # Built documentation output
+├── conf.py         # Sphinx configuration (mock imports, extensions, etc.)
 ├── index.rst       # Main documentation entry point
-├── requirements-base.txt    # Base documentation dependencies
-├── requirements-riva.txt    # Riva-specific dependencies
-├── requirements-playht.txt  # PlayHT-specific dependencies
 ├── build-docs.sh   # Local build script
-└── rtd-test.py     # ReadTheDocs test build script
+└── rtd-test.sh     # ReadTheDocs test build script (uses pip, not uv)
 ```
 
-## Notes
+## How It Works
 
-- Documentation is auto-generated from Python docstrings
-- Service modules are automatically detected and included
-- The build process matches our ReadTheDocs configuration
-- Warnings are treated as errors (-W flag) to maintain consistency
-- The --keep-going flag ensures all errors are reported
-- Dependencies are split into multiple requirements files to handle version conflicts
+- `conf.py` runs `sphinx-apidoc` during Sphinx's `setup()` phase to generate `.rst` files from Python source
+- Sphinx autodoc imports each module to extract docstrings
+- Modules with unavailable dependencies are listed in `autodoc_mock_imports` in `conf.py`
+- Napoleon extension converts Google-style docstrings to reStructuredText
 
 ## Troubleshooting
 
-If you encounter missing service modules:
+**Module not appearing in docs:**
 
-1. Verify the service is installed with its extras: `pip install pipecat-ai[service-name]`
-2. Check the build logs for import errors
-3. Ensure the service module is properly initialized in the package
-4. Run `./rtd-test.py` to test in an isolated environment matching ReadTheDocs
+1. Check the build output for `autodoc: failed to import` warnings
+2. If the module has an unresolvable import dependency, add it to `autodoc_mock_imports` in `conf.py`
+3. Verify the module is importable: `uv run python -c "import pipecat.module.name"`
 
-For dependency conflicts:
+**Duplicate object warnings:**
 
-1. Check the requirements files for version specifications
-2. Use `rtd-test.py` to verify dependency resolution
-3. Consider adding service-specific requirements files if needed
+These come from re-export modules or Sphinx discovering the same class through multiple import paths. Usually cosmetic.
 
-For more information:
+**Docstring formatting warnings:**
 
-- [ReadTheDocs Configuration](.readthedocs.yaml)
-- [Sphinx Documentation](https://www.sphinx-doc.org/)
+Docstrings use reStructuredText, not Markdown. Common issues:
+- Use `Example::` with indented code blocks, not `` ```python ``
+- Ensure blank lines between directive content and subsequent sections
+- Use `Parameters:` (not `Attributes:`) for dataclass field documentation to avoid duplicate entries

docs/api/build-docs.sh

@@ -1,8 +1,16 @@
 #!/bin/bash
 
+# Usage: ./build-docs.sh [--strict]
+#   --strict: Treat warnings as errors (default: warnings only)
+
+SPHINX_OPTS=""
+if [ "$1" = "--strict" ]; then
+    SPHINX_OPTS="-W --keep-going"
+fi
+
 # Build docs using uv
 echo "Installing dependencies with uv..."
-uv sync --group docs --all-extras --no-extra krisp --no-extra gstreamer --no-extra ultravox --no-extra local_smart_turn --no-extra moondream --no-extra riva --no-extra mlx-whisper
+uv sync --group docs --all-extras --no-extra gstreamer --no-extra local_smart_turn --no-extra moondream --no-extra mlx-whisper
 
 # Check if sphinx-build is available
 if ! uv run sphinx-build --version &> /dev/null; then
@@ -14,8 +22,7 @@ fi
 rm -rf _build
 
 echo "Building documentation..."
-# Build docs matching ReadTheDocs configuration
-uv run sphinx-build -b html -d _build/doctrees . _build/html -W --keep-going
+uv run sphinx-build -b html -d _build/doctrees . _build/html $SPHINX_OPTS
 
 if [ $? -eq 0 ]; then
     echo "Documentation built successfully!"
@@ -24,4 +31,4 @@ if [ $? -eq 0 ]; then
 else
     echo "Documentation build failed!" >&2
     exit 1
-fi
+fi

docs/api/conf.py

@@ -4,6 +4,19 @@ import sys
 from datetime import datetime
 from pathlib import Path
 
+# Fix Pydantic v2 + Sphinx autodoc incompatibility: ConfigDict(extra="allow") fails
+# during Sphinx's import because __pydantic_extra__ annotation on BaseModel resolves to
+# `Dict[str, Any] | None` whose get_origin() is Union, not dict. Patch the check to
+# accept Union-wrapped dict types (i.e., Optional[Dict[str, Any]]).
+import pydantic._internal._generate_schema as _pydantic_gs
+
+_ORIG_DICT_TYPES = _pydantic_gs.DICT_TYPES
+# Expand the accepted types to include Union (Optional[Dict[str, Any]])
+import types
+import typing
+
+_pydantic_gs.DICT_TYPES = [*_ORIG_DICT_TYPES, typing.Union, types.UnionType]
+
 # Configure logging
 logging.basicConfig(level=logging.INFO, format="%(asctime)s - %(levelname)s - %(message)s")
 logger = logging.getLogger("sphinx-build")
@@ -48,8 +61,6 @@ autodoc_default_options = {
 # Mock imports for optional dependencies
 autodoc_mock_imports = [
     # Krisp - has build issues on some platforms
-    "pipecat_ai_krisp",
-    "krisp",
     "krisp_audio",
     # System-specific GUI libraries
     "_tkinter",
@@ -61,9 +72,6 @@ autodoc_mock_imports = [
     # OpenCV - sometimes has import issues during docs build
     "cv2",
     # Heavy ML packages excluded from ReadTheDocs
-    # ultravox dependencies
-    "vllm",
-    "vllm.engine.arg_utils",
     # local-smart-turn dependencies
     "coremltools",
     "coremltools.models",
@@ -81,19 +89,29 @@ autodoc_mock_imports = [
     "einops",
     "intel_extension_for_pytorch",
     "huggingface_hub",
-    # riva dependencies
-    "riva",
-    "riva.client",
-    "riva.client.Auth",
-    "riva.client.ASRService",
-    "riva.client.StreamingRecognitionConfig",
-    "riva.client.RecognitionConfig",
-    "riva.client.AudioEncoding",
-    "riva.client.proto.riva_tts_pb2",
-    "riva.client.SpeechSynthesisService",
     # MLX dependencies (Apple Silicon specific)
     "mlx",
     "mlx_whisper",  # Note: might need underscore format too
+    # Pydantic v2 compatibility issues in third-party SDKs
+    "hume",
+    "hume.tts",
+    "hume.tts.types",
+    "cartesia",
+    "camb",
+    "sarvamai",
+    "openai.types.beta.realtime",
+    "langchain_core",
+    "langchain_core.messages",
+    # FastAPI - Pydantic v2 compatibility issues during Sphinx autodoc
+    "fastapi",
+    "fastapi.applications",
+    "fastapi.routing",
+    "fastapi.params",
+    "fastapi.middleware",
+    "fastapi.responses",
+    "uvicorn",
+    # Deepgram dependencies
+    "deepgram",
 ]
 
 # HTML output settings
@@ -119,8 +137,9 @@ def import_core_modules():
         "pipecat.observers",
         "pipecat.runner",
         "pipecat.serializers",
-        "pipecat.sync",
         "pipecat.transcriptions",
+        "pipecat.turns",
+        "pipecat.extensions",
         "pipecat.utils",
     ]
 
@@ -165,7 +184,6 @@ def setup(app):
     logger.info(f"Source directory: {source_dir}")
 
     excludes = [
-        str(project_root / "src/pipecat/pipeline/to_be_updated"),
         str(project_root / "src/pipecat/examples"),
         str(project_root / "src/pipecat/tests"),
         "**/test_*.py",

docs/api/index.rst

@@ -30,7 +30,7 @@ Quick Links
    Runner <api/pipecat.runner>
    Serializers <api/pipecat.serializers>
    Services <api/pipecat.services>
-   Sync <api/pipecat.sync>
    Transcriptions <api/pipecat.transcriptions>
    Transports <api/pipecat.transports>
-   Utils <api/pipecat.utils>
+   Turns <api/pipecat.turns>
+   Utils <api/pipecat.utils>

env.example

@@ -1,5 +1,5 @@
 # AI-COUSTICS
-AICOUSTICS_LICENSE_KEY=...
+AIC_LICENSE_KEY=...
 
 # Anthropic
 ANTHROPIC_API_KEY=...
@@ -31,6 +31,9 @@ AZURE_DALLE_API_KEY=...
 AZURE_DALLE_ENDPOINT=https://...
 AZURE_DALLE_MODEL=...
 
+# Camb.ai
+CAMB_API_KEY=...
+
 # Cartesia
 CARTESIA_API_KEY=...
 CARTESIA_VOICE_ID=...
@@ -40,11 +43,12 @@ CEREBRAS_API_KEY=...
 
 # Daily
 DAILY_API_KEY=...
-DAILY_SAMPLE_ROOM_URL=https://...
+DAILY_ROOM_URL=https://...
 
 # Deepgram
 DEEPGRAM_API_KEY=...
-SAGEMAKER_ENDPOINT_NAME=...
+SAGEMAKER_STT_ENDPOINT_NAME=...
+SAGEMAKER_TTS_ENDPOINT_NAME=...
 
 # DeepSeek
 DEEPSEEK_API_KEY=...
@@ -73,19 +77,23 @@ GOOGLE_CLOUD_PROJECT_ID=...
 GOOGLE_CLOUD_LOCATION=...
 GOOGLE_TEST_CREDENTIALS=...
 
-# Grok
-GROK_API_KEY=...
+# Gradium
+GRAPDIUM_API_KEY=...
 
 # Groq
 GROQ_API_KEY=...
 
 # Heygen
 HEYGEN_API_KEY=...
+HEYGEN_LIVE_AVATAR_API_KEY=...
 
 # Hume
 HUME_API_KEY=...
 HUME_VOICE_ID=...
 
+# Inception
+INCEPTION_API_KEY=...
+
 # Inworld
 INWORLD_API_KEY=...
 
@@ -93,7 +101,13 @@ INWORLD_API_KEY=...
 KRISP_MODEL_PATH=...
 
 # Krisp Viva
-KRISP_VIVA_MODEL_PATH=...
+KRISP_VIVA_API_KEY=...
+KRISP_VIVA_FILTER_MODEL_PATH=...
+KRISP_VIVA_TURN_MODEL_PATH=...
+
+# LemonSlice
+LEMONSLICE_API_KEY=...
+LEMONSLICE_AGENT_ID=...
 
 # LiveKit
 LIVEKIT_API_KEY=...
@@ -110,18 +124,25 @@ MINIMAX_GROUP_ID=...
 # Mistral
 MISTRAL_API_KEY=...
 
+# Nebius
+NEBIUS_API_KEY=...
+
 # Neuphonic
 NEUPHONIC_API_KEY=...
 
+# Novita
+NOVITA_API_KEY=...
+
 # NVIDIA
 NVIDIA_API_KEY=...
+# For a full example of how to deploy to SageMaker, see:
+# https://github.com/pipecat-ai/pipecat-examples/tree/main/nvidia_sagemaker_example/deployment/aws-sagemaker-nvidia
+SAGEMAKER_ASR_ENDPOINT_NAME=...
+SAGEMAKER_MAGPIE_ENDPOINT_NAME=...
 
 # OpenAI
 OPENAI_API_KEY=...
 
-# OpenPipe
-OPENPIPE_API_KEY=...
-
 # OpenRouter
 OPENROUTER_API_KEY=...
 
@@ -134,10 +155,6 @@ KOALA_ACCESS_KEY=...
 # Piper
 PIPER_BASE_URL=...
 
-# PlayHT
-PLAYHT_USER_ID=...
-PLAYHT_API_KEY=...
-
 # Plivo
 PLIVO_AUTH_ID=...
 PLIVO_AUTH_TOKEN=...
@@ -145,6 +162,10 @@ PLIVO_AUTH_TOKEN=...
 # Qwen
 QWEN_API_KEY=...
 
+# Resemble AI
+RESEMBLE_API_KEY=
+RESEMBLE_VOICE_UUID=
+
 # Rime
 RIME_API_KEY=...
 RIME_VOICE_ID=...
@@ -162,6 +183,9 @@ SENTRY_DSN=...
 SIMLI_API_KEY=...
 SIMLI_FACE_ID=...
 
+# Smallest
+SMALLEST_API_KEY=...
+
 # Smart turn
 LOCAL_SMART_TURN_MODEL_PATH=...
 FAL_SMART_TURN_API_KEY=...
@@ -187,8 +211,25 @@ TOGETHER_API_KEY=...
 TWILIO_ACCOUNT_SID=...
 TWILIO_AUTH_TOKEN=...
 
+# Ultravox Realtime
+ULTRAVOX_API_KEY=...
+
+# Vonage
+VONAGE_APPLICATION_ID=...
+VONAGE_SESSION_ID=...
+VONAGE_TOKEN=...
+
 # WhatsApp
 WHATSAPP_TOKEN=...
 WHATSAPP_WEBHOOK_VERIFICATION_TOKEN=...
 WHATSAPP_PHONE_NUMBER_ID=...
-WHATSAPP_APP_SECRET=...
+WHATSAPP_APP_SECRET=...
+
+# xAI / Grok
+XAI_API_KEY=...
+
+# PIPECAT_SCTP_MAX_CHUNK_SIZE controls the maximum SCTP DATA-chunk payload
+# size (bytes) used by aiortc's data channel. The default is 1100.
+# All the details here:
+# https://docs.pipecat.ai/api-reference/server/services/transport/small-webrtc#pipecat_sctp_max_chunk_size
+#PIPECAT_SCTP_MAX_CHUNK_SIZE=1100

examples/README.md

@@ -1,31 +1,150 @@
 # Pipecat Examples
 
-This directory contains examples to help you learn how to build with Pipecat.
+This directory contains examples showing how to build voice and multimodal agents with Pipecat.
 
-## Getting Started
+## Setup
 
-New to Pipecat? Start here:
+1. Follow the [README](https://github.com/pipecat-ai/pipecat/blob/main/README.md#%EF%B8%8F-contributing-to-the-framework) steps to get your local environment configured.
 
-- **[Quickstart](quickstart/)** - Get your first voice AI bot running in 5 minutes _(coming soon)_
-- **[Client/Server Web](client-server-web/)** - Learn to build web applications with Pipecat's client SDKs _(coming soon)_
-- **[Phone Bot with Twilio](phone-bot-twilio/)** - Connect your bot to a phone number _(coming soon)_
+   > **Run from root directory**: Make sure you are running the steps from the root directory.
 
-## Foundational Examples
+   > **Using local audio?**: The `LocalAudioTransport` requires a system dependency for `portaudio`. Install the dependency to use the transport.
 
-Single-file examples that introduce core Pipecat concepts one at a time. These examples:
+2. Copy the [`env.example`](../env.example) file and add API keys for services you plan to use:
 
-- Build on each other progressively
-- Focus on specific features or integrations
-- Are used for testing with every Pipecat release
+   ```bash
+   cp env.example .env
+   # Edit .env with your API keys
+   ```
 
-See the **[Foundational Examples README](foundational/)** for the complete list.
+3. Run any example:
 
-## More Advanced Examples
+   ```bash
+   uv run python getting-started/01-say-one-thing.py
+   ```
 
-Ready to explore complex use cases? Visit **[pipecat-examples](https://github.com/pipecat-ai/pipecat-examples)** for:
+4. Open the web interface at http://localhost:7860/client/ and click "Connect"
 
-- Production-ready applications
-- Multi-platform client implementations
-- Telephony integrations
-- Multimodal and creative applications
-- Deployment and monitoring examples
+## Running examples with other transports
+
+Most examples support running with other transports, like Twilio or Daily.
+
+### Daily
+
+You need to create a Daily account at https://dashboard.daily.co/u/signup. Once signed up, you can create your own room from the dashboard and set the environment variables `DAILY_ROOM_URL` and `DAILY_API_KEY`. Alternatively, you can let the example create a room for you (still needs `DAILY_API_KEY` environment variable). Then, start any example with `-t daily`:
+
+```bash
+uv run getting-started/06-voice-agent.py -t daily
+```
+
+### Twilio
+
+It is also possible to run the example through a Twilio phone number. You will need to setup a few things:
+
+1. Install and run [ngrok](https://ngrok.com/download).
+
+```bash
+ngrok http 7860
+```
+
+2. Configure your Twilio phone number. One way is to setup a TwiML app and set the request URL to the ngrok URL from step (1). Then, set your phone number to use the new TwiML app.
+
+Then, run the example with:
+
+```bash
+uv run getting-started/06-voice-agent.py -t twilio -x NGROK_HOST_NAME
+```
+
+## Directory Structure
+
+### [`getting-started/`](./getting-started/)
+
+Progressive introduction to Pipecat, from minimal TTS to a full voice agent with function calling.
+
+### [`voice/`](./voice/)
+
+Full STT + LLM + TTS voice agent pipelines showcasing different speech service providers (Deepgram, ElevenLabs, Cartesia, etc.)
+
+### [`function-calling/`](./function-calling/)
+
+Function calling with different LLM providers (OpenAI, Anthropic, Google, etc.)
+
+### [`transcription/`](./transcription/)
+
+Speech-to-text examples with various STT providers.
+
+### [`vision/`](./vision/)
+
+Image description and vision capabilities with different multimodal LLMs.
+
+### [`realtime/`](./realtime/)
+
+Realtime and multimodal live APIs (OpenAI Realtime, Gemini Live, AWS Nova Sonic, Ultravox, Grok).
+
+### [`persistent-context/`](./persistent-context/)
+
+Maintaining conversation context across sessions with different providers.
+
+### [`context-summarization/`](./context-summarization/)
+
+Summarizing conversation context to manage token limits.
+
+### [`update-settings/`](./update-settings/)
+
+Changing service settings at runtime, organized by service type:
+
+- **[`stt/`](./update-settings/stt/)** — Speech-to-text settings
+- **[`tts/`](./update-settings/tts/)** — Text-to-speech settings
+- **[`llm/`](./update-settings/llm/)** — LLM settings
+
+### [`turn-management/`](./turn-management/)
+
+Turn detection, interruption handling, and user input management.
+
+### [`thinking-and-mcp/`](./thinking-and-mcp/)
+
+LLM thinking/reasoning modes and MCP (Model Context Protocol) tool server integration.
+
+### [`transports/`](./transports/)
+
+Transport layer examples (WebRTC, Daily, LiveKit).
+
+### [`video-avatar/`](./video-avatar/)
+
+Video avatar integrations (Tavus, HeyGen, Simli, LemonSlice).
+
+### [`video-processing/`](./video-processing/)
+
+Video processing, mirroring, GStreamer, and custom video tracks.
+
+### [`audio/`](./audio/)
+
+Audio recording, background sounds, and sound effects.
+
+### [`observability/`](./observability/)
+
+Pipeline monitoring: observers, heartbeats, and Sentry metrics.
+
+### [`rag/`](./rag/)
+
+Retrieval-augmented generation, grounding, and long-term memory (Mem0, Gemini).
+
+### [`features/`](./features/)
+
+Miscellaneous features: wake phrases, live translation, service switching, voice switching, and more.
+
+## Advanced Usage
+
+### Customizing Network Settings
+
+```bash
+uv run python <example-name> --host 0.0.0.0 --port 8080
+```
+
+### Troubleshooting
+
+- **No audio/video**: Check browser permissions for microphone and camera
+- **Connection errors**: Verify API keys in `.env` file
+- **Port conflicts**: Use `--port` to change the port
+
+For more examples, visit the [pipecat-examples repository](https://github.com/pipecat-ai/pipecat-examples).

examples/audio/audio-bot-background-sound.py

@@ -1,5 +1,5 @@
 #
-# Copyright (c) 2024–2025, Daily
+# Copyright (c) 2024-2026, Daily
 #
 # SPDX-License-Identifier: BSD 2-Clause License
 #
@@ -12,16 +12,16 @@ from dotenv import load_dotenv
 from loguru import logger
 
 from pipecat.audio.mixers.soundfile_mixer import SoundfileMixer
-from pipecat.audio.turn.smart_turn.base_smart_turn import SmartTurnParams
-from pipecat.audio.turn.smart_turn.local_smart_turn_v3 import LocalSmartTurnAnalyzerV3
 from pipecat.audio.vad.silero import SileroVADAnalyzer
-from pipecat.audio.vad.vad_analyzer import VADParams
 from pipecat.frames.frames import LLMRunFrame, MixerEnableFrame, MixerUpdateSettingsFrame
 from pipecat.pipeline.pipeline import Pipeline
 from pipecat.pipeline.runner import PipelineRunner
 from pipecat.pipeline.task import PipelineParams, PipelineTask
 from pipecat.processors.aggregators.llm_context import LLMContext
-from pipecat.processors.aggregators.llm_response_universal import LLMContextAggregatorPair
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
 from pipecat.runner.types import RunnerArguments
 from pipecat.runner.utils import create_transport
 from pipecat.services.cartesia.tts import CartesiaTTSService
@@ -34,12 +34,11 @@ from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
 load_dotenv(override=True)
 
 OFFICE_SOUND_FILE = os.path.join(
-    os.path.dirname(__file__), "assets", "office-ambience-24000-mono.mp3"
+    os.path.dirname(__file__), "../assets", "office-ambience-24000-mono.mp3"
 )
 
-# We store functions so objects (e.g. SileroVADAnalyzer) don't get
-# instantiated. The function will be called when the desired transport gets
-# selected.
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
 transport_params = {
     "daily": lambda: DailyParams(
         audio_in_enabled=True,
@@ -49,8 +48,6 @@ transport_params = {
             default_sound="office",
             volume=2.0,
         ),
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "twilio": lambda: FastAPIWebsocketParams(
         audio_in_enabled=True,
@@ -60,8 +57,6 @@ transport_params = {
             default_sound="office",
             volume=2.0,
         ),
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "webrtc": lambda: TransportParams(
         audio_in_enabled=True,
@@ -71,41 +66,42 @@ transport_params = {
             default_sound="office",
             volume=2.0,
         ),
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
 }
 
 
 async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
-    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+    stt = DeepgramSTTService(api_key=os.environ["DEEPGRAM_API_KEY"])
 
     tts = CartesiaTTSService(
-        api_key=os.getenv("CARTESIA_API_KEY"),
-        voice_id="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        api_key=os.environ["CARTESIA_API_KEY"],
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
     )
 
-    llm = OpenAILLMService(api_key=os.getenv("OPENAI_API_KEY"))
+    llm = OpenAILLMService(
+        api_key=os.environ["OPENAI_API_KEY"],
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way.",
+        ),
+    )
 
-    messages = [
-        {
-            "role": "system",
-            "content": "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be spoken aloud, so avoid special characters that can't easily be spoken, such as emojis or bullet points. Respond to what the user said in a creative and helpful way.",
-        },
-    ]
-
-    context = LLMContext(messages)
-    context_aggregator = LLMContextAggregatorPair(context)
+    context = LLMContext()
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
 
     pipeline = Pipeline(
         [
             transport.input(),  # Transport user input
             stt,  # STT
-            context_aggregator.user(),  # User responses
+            user_aggregator,  # User responses
             llm,  # LLM
             tts,  # TTS
             transport.output(),  # Transport bot output
-            context_aggregator.assistant(),  # Assistant spoken responses
+            assistant_aggregator,  # Assistant spoken responses
         ]
     )
 
@@ -132,7 +128,9 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         logger.info(f"Re-enabling background sound and starting bot...")
         await task.queue_frame(MixerEnableFrame(True))
         # Kick off the conversation.
-        messages.append({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/audio/audio-recording.py

@@ -1,5 +1,5 @@
 #
-# Copyright (c) 2024–2025, Daily
+# Copyright (c) 2024-2026, Daily
 #
 # SPDX-License-Identifier: BSD 2-Clause License
 #
@@ -50,16 +50,16 @@ import aiofiles
 from dotenv import load_dotenv
 from loguru import logger
 
-from pipecat.audio.turn.smart_turn.base_smart_turn import SmartTurnParams
-from pipecat.audio.turn.smart_turn.local_smart_turn_v3 import LocalSmartTurnAnalyzerV3
 from pipecat.audio.vad.silero import SileroVADAnalyzer
-from pipecat.audio.vad.vad_analyzer import VADParams
 from pipecat.frames.frames import LLMRunFrame
 from pipecat.pipeline.pipeline import Pipeline
 from pipecat.pipeline.runner import PipelineRunner
 from pipecat.pipeline.task import PipelineParams, PipelineTask
 from pipecat.processors.aggregators.llm_context import LLMContext
-from pipecat.processors.aggregators.llm_response_universal import LLMContextAggregatorPair
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
 from pipecat.processors.audio.audio_buffer_processor import AudioBufferProcessor
 from pipecat.runner.types import RunnerArguments
 from pipecat.runner.utils import create_transport
@@ -87,27 +87,20 @@ async def save_audio_file(audio: bytes, filename: str, sample_rate: int, num_cha
         logger.info(f"Audio saved to {filename}")
 
 
-# We store functions so objects (e.g. SileroVADAnalyzer) don't get
-# instantiated. The function will be called when the desired transport gets
-# selected.
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
 transport_params = {
     "daily": lambda: DailyParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "twilio": lambda: FastAPIWebsocketParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "webrtc": lambda: TransportParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
 }
 
@@ -115,38 +108,41 @@ transport_params = {
 async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     logger.info(f"Starting bot")
 
-    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"), audio_passthrough=True)
+    stt = DeepgramSTTService(api_key=os.environ["DEEPGRAM_API_KEY"], audio_passthrough=True)
 
     tts = CartesiaTTSService(
-        api_key=os.getenv("CARTESIA_API_KEY"),
-        voice_id="71a7ad14-091c-4e8e-a314-022ece01c121",
+        api_key=os.environ["CARTESIA_API_KEY"],
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",
+        ),
     )
 
-    llm = OpenAILLMService(api_key=os.getenv("OPENAI_API_KEY"), model="gpt-4")
+    llm = OpenAILLMService(
+        api_key=os.environ["OPENAI_API_KEY"],
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way.",
+        ),
+    )
 
     # Create audio buffer processor
     audiobuffer = AudioBufferProcessor()
 
-    messages = [
-        {
-            "role": "system",
-            "content": "You are a helpful assistant demonstrating audio recording capabilities. Keep your responses brief and clear.",
-        },
-    ]
-
-    context = LLMContext(messages)
-    context_aggregator = LLMContextAggregatorPair(context)
+    context = LLMContext()
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
 
     pipeline = Pipeline(
         [
             transport.input(),
             stt,
-            context_aggregator.user(),
+            user_aggregator,
             llm,
             tts,
             transport.output(),
             audiobuffer,  # Add audio buffer to pipeline
-            context_aggregator.assistant(),
+            assistant_aggregator,
         ]
     )
 

examples/audio/audio-sound-effects.py

@@ -1,5 +1,5 @@
 #
-# Copyright (c) 2024–2025, Daily
+# Copyright (c) 2024-2026, Daily
 #
 # SPDX-License-Identifier: BSD 2-Clause License
 #
@@ -10,10 +10,7 @@ import wave
 from dotenv import load_dotenv
 from loguru import logger
 
-from pipecat.audio.turn.smart_turn.base_smart_turn import SmartTurnParams
-from pipecat.audio.turn.smart_turn.local_smart_turn_v3 import LocalSmartTurnAnalyzerV3
 from pipecat.audio.vad.silero import SileroVADAnalyzer
-from pipecat.audio.vad.vad_analyzer import VADParams
 from pipecat.frames.frames import (
     Frame,
     LLMContextFrame,
@@ -25,7 +22,10 @@ from pipecat.pipeline.pipeline import Pipeline
 from pipecat.pipeline.runner import PipelineRunner
 from pipecat.pipeline.task import PipelineTask
 from pipecat.processors.aggregators.llm_context import LLMContext
-from pipecat.processors.aggregators.llm_response_universal import LLMContextAggregatorPair
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
 from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
 from pipecat.processors.logger import FrameLogger
 from pipecat.runner.types import RunnerArguments
@@ -81,27 +81,20 @@ class InboundSoundEffectWrapper(FrameProcessor):
             await self.push_frame(frame, direction)
 
 
-# We store functions so objects (e.g. SileroVADAnalyzer) don't get
-# instantiated. The function will be called when the desired transport gets
-# selected.
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
 transport_params = {
     "daily": lambda: DailyParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "twilio": lambda: FastAPIWebsocketParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "webrtc": lambda: TransportParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
 }
 
@@ -109,24 +102,27 @@ transport_params = {
 async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     logger.info(f"Starting bot")
 
-    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+    stt = DeepgramSTTService(api_key=os.environ["DEEPGRAM_API_KEY"])
 
-    llm = OpenAILLMService(api_key=os.getenv("OPENAI_API_KEY"))
-
-    tts = CartesiaTTSService(
-        api_key=os.getenv("CARTESIA_API_KEY"),
-        voice_id="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+    llm = OpenAILLMService(
+        api_key=os.environ["OPENAI_API_KEY"],
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way.",
+        ),
     )
 
-    messages = [
-        {
-            "role": "system",
-            "content": "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be spoken aloud, so avoid special characters that can't easily be spoken, such as emojis or bullet points. Respond to what the user said in a creative and helpful way.",
-        },
-    ]
+    tts = CartesiaTTSService(
+        api_key=os.environ["CARTESIA_API_KEY"],
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
+    )
 
-    context = LLMContext(messages)
-    context_aggregator = LLMContextAggregatorPair(context)
+    context = LLMContext()
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
     out_sound = OutboundSoundEffectWrapper()
     in_sound = InboundSoundEffectWrapper()
     fl = FrameLogger("LLM Out")
@@ -136,7 +132,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         [
             transport.input(),
             stt,
-            context_aggregator.user(),
+            user_aggregator,
             in_sound,
             fl2,
             llm,
@@ -144,7 +140,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
             tts,
             out_sound,
             transport.output(),
-            context_aggregator.assistant(),
+            assistant_aggregator,
         ]
     )
 

examples/context-summarization/context-summarization-dedicated-llm.py

@@ -0,0 +1,238 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+"""Example demonstrating advanced context summarization configuration.
+
+This example shows how to customize context summarization with:
+- A dedicated cheap/fast LLM for generating summaries (Gemini Flash)
+- A custom summary message template (XML tags)
+- A custom summarization prompt
+- A summarization timeout
+- The on_summary_applied event for observability
+"""
+
+import asyncio
+import os
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.adapters.schemas.function_schema import FunctionSchema
+from pipecat.adapters.schemas.tools_schema import ToolsSchema
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import LLMRunFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_context_summarizer import SummaryAppliedEvent
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMAssistantAggregatorParams,
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.google.llm import GoogleLLMService
+from pipecat.services.llm_service import FunctionCallParams
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+from pipecat.utils.context.llm_context_summarization import (
+    LLMAutoContextSummarizationConfig,
+    LLMContextSummaryConfig,
+)
+
+load_dotenv(override=True)
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+# Custom summarization prompt tailored to the application
+CUSTOM_SUMMARIZATION_PROMPT = """Summarize this conversation, preserving:
+- Key decisions and agreements
+- Important facts and user preferences
+- Any pending action items or unresolved questions
+
+Be concise. Use clear, factual statements grouped by topic.
+Omit greetings, small talk, and resolved tangents."""
+
+
+# Tool functions for the LLM
+async def get_current_weather(params: FunctionCallParams):
+    """Get the current weather."""
+    logger.info("Tool called: get_current_weather")
+    await asyncio.sleep(1)  # Simulate some processing
+    await params.result_callback({"conditions": "nice", "temperature": "75"})
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info("Starting bot")
+
+    stt = DeepgramSTTService(api_key=os.environ["DEEPGRAM_API_KEY"])
+
+    tts = CartesiaTTSService(
+        api_key=os.environ["CARTESIA_API_KEY"],
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
+    )
+
+    system_prompt = """You are a helpful LLM in a voice call. Your goal is to demonstrate your
+                    capabilities in a succinct way. Your output will be spoken aloud, so avoid
+                    special characters that can't easily be spoken, such as emojis or bullet points.
+                    Respond to what the user said in a creative and helpful way.
+                    You have access to tools to get the current weather - use them when relevant.
+                    When you see a <context_summary> block, it contains a compressed summary
+                    of earlier conversation. Use it as reference but don't mention it to the user.
+                    """
+
+    # Primary LLM for conversation (could be any provider)
+    llm = OpenAILLMService(
+        api_key=os.environ["OPENAI_API_KEY"],
+        settings=OpenAILLMService.Settings(
+            system_instruction=system_prompt,
+        ),
+    )
+
+    # Dedicated cheap/fast LLM for summarization only
+    summarization_llm = GoogleLLMService(
+        api_key=os.environ["GOOGLE_API_KEY"],
+        settings=GoogleLLMService.Settings(
+            model="gemini-2.5-flash",
+        ),
+    )
+
+    # Register tool functions
+    llm.register_function("get_current_weather", get_current_weather)
+
+    weather_function = FunctionSchema(
+        name="get_current_weather",
+        description="Get the current weather",
+        properties={
+            "location": {
+                "type": "string",
+                "description": "The city and state, e.g. San Francisco, CA",
+            },
+            "format": {
+                "type": "string",
+                "enum": ["celsius", "fahrenheit"],
+                "description": "The temperature unit to use. Infer this from the user's location.",
+            },
+        },
+        required=["location", "format"],
+    )
+    tools = ToolsSchema(standard_tools=[weather_function])
+
+    context = LLMContext(tools=tools)
+
+    # Create aggregators with custom summarization
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(
+            vad_analyzer=SileroVADAnalyzer(),
+        ),
+        assistant_params=LLMAssistantAggregatorParams(
+            enable_auto_context_summarization=True,
+            auto_context_summarization_config=LLMAutoContextSummarizationConfig(
+                # Trigger thresholds (low values to demonstrate quickly)
+                max_context_tokens=1000,
+                max_unsummarized_messages=10,
+                summary_config=LLMContextSummaryConfig(
+                    # Summary generation
+                    target_context_tokens=800,
+                    min_messages_after_summary=2,
+                    summarization_prompt=CUSTOM_SUMMARIZATION_PROMPT,
+                    # Custom summary format - wrap in XML tags so the system
+                    # prompt can identify summaries vs. live conversation
+                    summary_message_template="<context_summary>\n{summary}\n</context_summary>",
+                    # Use a dedicated cheap LLM for summarization instead of
+                    # the primary conversation model
+                    llm=summarization_llm,
+                    # Cancel summarization if it takes longer than 60 seconds
+                    summarization_timeout=60.0,
+                ),
+            ),
+        ),
+    )
+
+    # Listen for summarization events
+    @assistant_aggregator.event_handler("on_summary_applied")
+    async def on_summary_applied(aggregator, summarizer, event: SummaryAppliedEvent):
+        logger.info(
+            f"Context summarized: {event.original_message_count} messages -> "
+            f"{event.new_message_count} messages "
+            f"({event.summarized_message_count} summarized, "
+            f"{event.preserved_message_count} preserved)"
+        )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),  # Transport user input
+            stt,
+            user_aggregator,  # User responses
+            llm,  # LLM
+            tts,  # TTS
+            transport.output(),  # Transport bot output
+            assistant_aggregator,  # Assistant spoken responses
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info("Client connected")
+        # Kick off the conversation.
+        context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info("Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/context-summarization/context-summarization-google.py

@@ -0,0 +1,199 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+"""Example demonstrating context summarization feature.
+
+This example shows how to enable and configure context summarization to automatically
+compress conversation history when token limits are approached. It also demonstrates
+that summarization correctly handles function calls, preserving incomplete function
+call sequences.
+"""
+
+import asyncio
+import os
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.adapters.schemas.function_schema import FunctionSchema
+from pipecat.adapters.schemas.tools_schema import ToolsSchema
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import LLMRunFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_context_summarizer import SummaryAppliedEvent
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMAssistantAggregatorParams,
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.google.llm import GoogleLLMService
+from pipecat.services.llm_service import FunctionCallParams
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+from pipecat.utils.context.llm_context_summarization import (
+    LLMAutoContextSummarizationConfig,
+    LLMContextSummaryConfig,
+)
+
+load_dotenv(override=True)
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+# Tool functions for the LLM
+async def get_current_weather(params: FunctionCallParams):
+    """Get the current time in a readable format."""
+    logger.info("Tool called: get_current_weather")
+    await asyncio.sleep(1)  # Simulate some processing
+    await params.result_callback({"conditions": "nice", "temperature": "75"})
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info("Starting bot")
+
+    stt = DeepgramSTTService(api_key=os.environ["DEEPGRAM_API_KEY"])
+
+    tts = CartesiaTTSService(
+        api_key=os.environ["CARTESIA_API_KEY"],
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
+    )
+
+    llm = GoogleLLMService(
+        api_key=os.environ["GOOGLE_API_KEY"],
+        settings=GoogleLLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way. You have access to tools to get the current weather - use them when relevant.",
+        ),
+    )
+
+    # Register tool functions
+    llm.register_function("get_current_weather", get_current_weather)
+
+    weather_function = FunctionSchema(
+        name="get_current_weather",
+        description="Get the current weather",
+        properties={
+            "location": {
+                "type": "string",
+                "description": "The city and state, e.g. San Francisco, CA",
+            },
+            "format": {
+                "type": "string",
+                "enum": ["celsius", "fahrenheit"],
+                "description": "The temperature unit to use. Infer this from the user's location.",
+            },
+        },
+        required=["location", "format"],
+    )
+    tools = ToolsSchema(standard_tools=[weather_function])
+
+    context = LLMContext(tools=tools)
+
+    # Create aggregators with summarization enabled
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(
+            vad_analyzer=SileroVADAnalyzer(),
+        ),
+        assistant_params=LLMAssistantAggregatorParams(
+            enable_auto_context_summarization=True,
+            # Optional: customize context summarization behavior
+            # Using low limits to demonstrate the feature quickly
+            auto_context_summarization_config=LLMAutoContextSummarizationConfig(
+                max_context_tokens=1000,  # Trigger summarization at 1000 tokens
+                max_unsummarized_messages=10,  # Or when 10 new messages accumulate
+                summary_config=LLMContextSummaryConfig(
+                    target_context_tokens=800,  # Target context size for the summarization
+                    min_messages_after_summary=2,  # Keep last 2 messages uncompressed
+                ),
+            ),
+        ),
+    )
+
+    # Listen for summarization events
+    @assistant_aggregator.event_handler("on_summary_applied")
+    async def on_summary_applied(aggregator, summarizer, event: SummaryAppliedEvent):
+        logger.info(
+            f"Context summarized: {event.original_message_count} messages -> "
+            f"{event.new_message_count} messages "
+            f"({event.summarized_message_count} summarized, "
+            f"{event.preserved_message_count} preserved)"
+        )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),  # Transport user input
+            stt,
+            user_aggregator,  # User responses
+            llm,  # LLM
+            tts,  # TTS
+            transport.output(),  # Transport bot output
+            assistant_aggregator,  # Assistant spoken responses
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info("Client connected")
+        # Kick off the conversation.
+        context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info("Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/context-summarization/context-summarization-manual-openai.py

@@ -0,0 +1,173 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+"""Example demonstrating manual context summarization via a function call.
+
+This example shows how to trigger context summarization on demand rather than
+automatically. The user can ask the bot to "summarize the conversation" and the
+bot will call a function that pushes an LLMSummarizeContextFrame into the
+pipeline, causing the LLM service to compress the conversation history.
+
+Unlike example 54, automatic summarization is NOT enabled here. Summarization
+only happens when the user explicitly requests it through the function call.
+"""
+
+import os
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.adapters.schemas.function_schema import FunctionSchema
+from pipecat.adapters.schemas.tools_schema import ToolsSchema
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import LLMRunFrame, LLMSummarizeContextFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.llm_service import FunctionCallParams
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+
+load_dotenv(override=True)
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def summarize_conversation(params: FunctionCallParams):
+    """Trigger manual context summarization via a pipeline frame."""
+    logger.info("Tool called: summarize_conversation")
+    await params.result_callback({"status": "summarization_requested"})
+    await params.llm.queue_frame(LLMSummarizeContextFrame())
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info("Starting bot")
+
+    stt = DeepgramSTTService(api_key=os.environ["DEEPGRAM_API_KEY"])
+
+    tts = CartesiaTTSService(
+        api_key=os.environ["CARTESIA_API_KEY"],
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
+    )
+
+    system_prompt = """You are a helpful LLM in a voice call. Your goal is to demonstrate your
+                    capabilities in a succinct way. Your output will be spoken aloud, so avoid
+                    special characters that can't easily be spoken, such as emojis or bullet points.
+                    Respond to what the user said in a creative and helpful way.
+                    If the user asks you to summarize the conversation, call the
+                    summarize_conversation function. After summarization, briefly acknowledge
+                    that the conversation history has been compressed.
+                    """
+
+    llm = OpenAILLMService(
+        api_key=os.environ["OPENAI_API_KEY"],
+        settings=OpenAILLMService.Settings(
+            system_instruction=system_prompt,
+        ),
+    )
+
+    llm.register_function("summarize_conversation", summarize_conversation)
+
+    summarize_function = FunctionSchema(
+        name="summarize_conversation",
+        description=(
+            "Summarize and compress the conversation history. "
+            "Call this when the user asks you to summarize the conversation "
+            "or when you want to free up context space."
+        ),
+        properties={},
+        required=[],
+    )
+    tools = ToolsSchema(standard_tools=[summarize_function])
+
+    context = LLMContext(tools=tools)
+
+    # Automatic summarization is NOT enabled here (enable_auto_context_summarization
+    # defaults to False). The summarizer is still created internally so that
+    # LLMSummarizeContextFrame frames pushed via the function call are handled.
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),  # Transport user input
+            stt,
+            user_aggregator,  # User responses
+            llm,  # LLM
+            tts,  # TTS
+            transport.output(),  # Transport bot output
+            assistant_aggregator,  # Assistant spoken responses
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info("Client connected")
+        # Kick off the conversation.
+        context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info("Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/context-summarization/context-summarization-openai.py

@@ -0,0 +1,199 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+"""Example demonstrating context summarization feature.
+
+This example shows how to enable and configure context summarization to automatically
+compress conversation history when token limits are approached. It also demonstrates
+that summarization correctly handles function calls, preserving incomplete function
+call sequences.
+"""
+
+import asyncio
+import os
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.adapters.schemas.function_schema import FunctionSchema
+from pipecat.adapters.schemas.tools_schema import ToolsSchema
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import LLMRunFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_context_summarizer import SummaryAppliedEvent
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMAssistantAggregatorParams,
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.llm_service import FunctionCallParams
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+from pipecat.utils.context.llm_context_summarization import (
+    LLMAutoContextSummarizationConfig,
+    LLMContextSummaryConfig,
+)
+
+load_dotenv(override=True)
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+# Tool functions for the LLM
+async def get_current_weather(params: FunctionCallParams):
+    """Get the current time in a readable format."""
+    logger.info("Tool called: get_current_weather")
+    await asyncio.sleep(1)  # Simulate some processing
+    await params.result_callback({"conditions": "nice", "temperature": "75"})
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info("Starting bot")
+
+    stt = DeepgramSTTService(api_key=os.environ["DEEPGRAM_API_KEY"])
+
+    tts = CartesiaTTSService(
+        api_key=os.environ["CARTESIA_API_KEY"],
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
+    )
+
+    llm = OpenAILLMService(
+        api_key=os.environ["OPENAI_API_KEY"],
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way. You have access to tools to get the current weather - use them when relevant.",
+        ),
+    )
+
+    # Register tool functions
+    llm.register_function("get_current_weather", get_current_weather)
+
+    weather_function = FunctionSchema(
+        name="get_current_weather",
+        description="Get the current weather",
+        properties={
+            "location": {
+                "type": "string",
+                "description": "The city and state, e.g. San Francisco, CA",
+            },
+            "format": {
+                "type": "string",
+                "enum": ["celsius", "fahrenheit"],
+                "description": "The temperature unit to use. Infer this from the user's location.",
+            },
+        },
+        required=["location", "format"],
+    )
+    tools = ToolsSchema(standard_tools=[weather_function])
+
+    context = LLMContext(tools=tools)
+
+    # Create aggregators with summarization enabled
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(
+            vad_analyzer=SileroVADAnalyzer(),
+        ),
+        assistant_params=LLMAssistantAggregatorParams(
+            enable_auto_context_summarization=True,
+            # Optional: customize context summarization behavior
+            # Using low limits to demonstrate the feature quickly
+            auto_context_summarization_config=LLMAutoContextSummarizationConfig(
+                max_context_tokens=1000,  # Trigger summarization at 1000 tokens
+                max_unsummarized_messages=10,  # Or when 10 new messages accumulate
+                summary_config=LLMContextSummaryConfig(
+                    target_context_tokens=800,  # Target context size for the summarization
+                    min_messages_after_summary=2,  # Keep last 2 messages uncompressed
+                ),
+            ),
+        ),
+    )
+
+    # Listen for summarization events
+    @assistant_aggregator.event_handler("on_summary_applied")
+    async def on_summary_applied(aggregator, summarizer, event: SummaryAppliedEvent):
+        logger.info(
+            f"Context summarized: {event.original_message_count} messages -> "
+            f"{event.new_message_count} messages "
+            f"({event.summarized_message_count} summarized, "
+            f"{event.preserved_message_count} preserved)"
+        )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),  # Transport user input
+            stt,
+            user_aggregator,  # User responses
+            llm,  # LLM
+            tts,  # TTS
+            transport.output(),  # Transport bot output
+            assistant_aggregator,  # Assistant spoken responses
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info("Client connected")
+        # Kick off the conversation.
+        context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info("Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/features-add-tool-change-messages.py

@@ -0,0 +1,232 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+"""Manual validation harness for the ``add_tool_change_messages`` feature.
+
+When tools change mid-conversation, LLMs can produce a few different
+flavors of tool-call-related hallucination:
+
+- **Forward hallucination** — calling a tool that has been removed.
+- **Negative hallucination** — refusing to call a tool that has been
+  re-added (because recent context is full of "I can't" responses).
+- **Hallucinated output when tools are unavailable** — making up an
+  answer rather than declining gracefully, or producing JSON that
+  *looks* like a tool call but is actually just an assistant text
+  response.
+
+The ``add_tool_change_messages`` feature mitigates these by appending a
+developer-role message to the conversation whenever ``LLMSetToolsFrame``
+changes the set of advertised tools, so the LLM stays in sync with what's
+actually available.
+
+This harness exercises all of those flavors by flipping the advertised
+tool set on a turn counter:
+
+    Phase 0 (turns 1–4):   weather tool ACTIVE — confirm baseline.
+    Phase 1 (turns 5–8):   tool REMOVED — keep asking for weather.
+    Phase 2 (turn 9+):     tool RE-ADDED — does the LLM call it again?
+
+Set ``ADD_TOOL_CHANGE_MESSAGES=0`` to disable the mitigation and see the
+unmitigated behavior. The default is ON so a fresh run shows the feature
+working.
+
+Defaults to Llama 3.1 8B Instruct via a locally-running Ollama —
+anecdotally one of the more hallucination-prone of the easily accessible
+models. Pull the model once with ``ollama pull llama3.1:8b`` and make
+sure ``ollama serve`` is running. Swap the LLM service to validate other
+providers.
+
+Run with::
+
+    uv run examples/features/features-add-tool-change-messages.py
+    ADD_TOOL_CHANGE_MESSAGES=0 uv run examples/features/features-add-tool-change-messages.py
+"""
+
+import os
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.adapters.schemas.function_schema import FunctionSchema
+from pipecat.adapters.schemas.tools_schema import ToolsSchema
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import LLMRunFrame, LLMSetToolsFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import NOT_GIVEN, LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.llm_service import FunctionCallParams
+from pipecat.services.ollama.llm import OLLamaLLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+
+load_dotenv(override=True)
+
+# Default ON so a fresh run shows the feature working. Set to "0" to A/B
+# against the unmitigated behavior.
+ADD_TOOL_CHANGE_MESSAGES = os.environ.get("ADD_TOOL_CHANGE_MESSAGES", "1") == "1"
+
+
+async def fetch_weather_from_api(params: FunctionCallParams):
+    await params.result_callback({"conditions": "nice", "temperature": "75"})
+
+
+weather_function = FunctionSchema(
+    name="get_current_weather",
+    description="Get the current weather",
+    properties={
+        "location": {
+            "type": "string",
+            "description": "The city and state, e.g. San Francisco, CA",
+        },
+        "format": {
+            "type": "string",
+            "enum": ["celsius", "fahrenheit"],
+            "description": "The temperature unit to use. Infer this from the user's location.",
+        },
+    },
+    required=["location", "format"],
+)
+weather_tools = ToolsSchema(standard_tools=[weather_function])
+
+
+transport_params = {
+    "daily": lambda: DailyParams(audio_in_enabled=True, audio_out_enabled=True),
+    "twilio": lambda: FastAPIWebsocketParams(audio_in_enabled=True, audio_out_enabled=True),
+    "webrtc": lambda: TransportParams(audio_in_enabled=True, audio_out_enabled=True),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(
+        f"Starting add_tool_change_messages demo bot "
+        f"(ADD_TOOL_CHANGE_MESSAGES={ADD_TOOL_CHANGE_MESSAGES})"
+    )
+
+    stt = DeepgramSTTService(api_key=os.environ["DEEPGRAM_API_KEY"])
+
+    tts = CartesiaTTSService(
+        api_key=os.environ["CARTESIA_API_KEY"],
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
+    )
+
+    llm = OLLamaLLMService(
+        settings=OLLamaLLMService.Settings(
+            # Llama 3.1 8B Instruct is anecdotally one of the more
+            # hallucination-prone of the easily accessible models — exactly
+            # what we want for this validation harness. Pull it with
+            # ``ollama pull llama3.1:8b`` and make sure ``ollama serve``
+            # is running.
+            model="llama3.1:8b",
+            system_instruction=(
+                "You are a helpful assistant in a voice conversation. Your responses "
+                "will be spoken aloud, so avoid emojis, bullet points, or other "
+                "formatting that can't be spoken. Respond briefly and naturally. "
+                "If the user asks for the current weather, use the `get_current_weather` "
+                "function if it's available. IMPORTANT: if you do not have access to the function, "
+                "say something along the lines of 'Sorry, I can't check the weather right now.'."
+            ),
+        ),
+    )
+    llm.register_function("get_current_weather", fetch_weather_from_api)
+
+    context = LLMContext(tools=weather_tools)
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+        add_tool_change_messages=ADD_TOOL_CHANGE_MESSAGES,
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),
+            stt,
+            user_aggregator,
+            llm,
+            tts,
+            transport.output(),
+            assistant_aggregator,
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(enable_metrics=True, enable_usage_metrics=True),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    # Phase controller: roughly 4 turns per phase.
+    user_turn_count = 0
+    REMOVE_AT_TURN = 5  # tool gone for turn N onward
+    READD_AT_TURN = 9  # tool back for turn N onward
+
+    @user_aggregator.event_handler("on_user_turn_stopped")
+    async def on_user_turn_stopped(aggregator, strategy, message):
+        nonlocal user_turn_count
+        user_turn_count += 1
+        logger.info(f"=== User turn {user_turn_count} complete ===")
+
+        if user_turn_count == REMOVE_AT_TURN - 1:
+            logger.info(
+                "=== Phase 1: weather tool REMOVED. Keep asking about the weather "
+                "to exercise hallucination scenarios. ==="
+            )
+            await task.queue_frame(LLMSetToolsFrame(tools=NOT_GIVEN))
+        elif user_turn_count == READD_AT_TURN - 1:
+            logger.info(
+                "=== Phase 2: weather tool RE-ADDED. Ask for the weather again — "
+                "does the LLM call it, or keep refusing? (THIS IS THE TEST.) ==="
+            )
+            await task.queue_frame(LLMSetToolsFrame(tools=weather_tools))
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info("Client connected")
+        logger.info(
+            "=== Phase 0: weather tool ACTIVE. Ask for the weather a few times "
+            "to confirm it's working. ==="
+        )
+        context.add_message(
+            {
+                "role": "developer",
+                "content": (
+                    "Please introduce yourself briefly to the user, then invite them "
+                    "to ask about the weather."
+                ),
+            }
+        )
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info("Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/features-app-resources.py

@@ -0,0 +1,327 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+"""Example demonstrating ``PipelineTask(app_resources=...)``.
+
+``app_resources`` is an application-defined bag of anything your
+application code may want to share across a session: database handles,
+HTTP clients, feature flags, per-user state, observability clients,
+in-memory caches — whatever fits your app. Pipecat passes it through
+untouched and exposes it as ``task.app_resources``, so any code with a
+handle on the task can read or mutate it.
+
+Two of the convenience aliases exercised below:
+
+- Tool handlers read it from ``FunctionCallParams.app_resources``.
+- Custom ``FrameProcessor`` subclasses read it from
+  ``self.pipeline_task.app_resources``.
+
+This example uses two small loggers as stand-ins for that "shared thing":
+``ToolCallLogger`` (written from tool handlers) and
+``TranscriptionLogger`` (written from a custom ``FrameProcessor`` that
+sits in the pipeline). A real app might just as easily pass a Postgres
+pool, a Redis client, a Stripe SDK instance, or any combination thereof.
+The mechanics shown here — construct once, hand to the task, read it
+from each site, inspect it after the session — are the same regardless
+of what you put in.
+
+We bundle resources in a typed ``AppResources`` dataclass and cast back
+to it at each read site. Pipecat doesn't care what type you pass (a
+plain dict works too), but a typed container gives you autocomplete and
+refactor safety instead of dict-by-string-key lookups.
+"""
+
+import json
+import os
+from collections.abc import Mapping
+from dataclasses import dataclass
+from datetime import UTC, datetime
+from typing import Any, cast
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.adapters.schemas.function_schema import FunctionSchema
+from pipecat.adapters.schemas.tools_schema import ToolsSchema
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import Frame, LLMRunFrame, TranscriptionFrame, TTSSpeakFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.llm_service import FunctionCallParams
+from pipecat.services.openai.responses.llm import OpenAIResponsesLLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+
+load_dotenv(override=True)
+
+
+class ToolCallLogger:
+    """Stand-in shared resource — swap for whatever your app actually needs."""
+
+    def __init__(self):
+        """Initialize the logger with an empty list of recorded calls."""
+        self._calls: list[dict[str, Any]] = []
+
+    def log_tool_call(self, function_name: str, arguments: Mapping[str, Any]) -> None:
+        """Record a tool call invocation.
+
+        Args:
+            function_name: The name of the tool being invoked.
+            arguments: The arguments passed to the tool.
+        """
+        entry = {
+            "timestamp": datetime.now(UTC).isoformat(),
+            "function_name": function_name,
+            "arguments": dict(arguments),
+        }
+        self._calls.append(entry)
+        logger.info(f"[ToolCallLogger] {function_name} called with {dict(arguments)}")
+
+    def dump(self) -> str:
+        """Return all recorded tool calls as a JSON string."""
+        return json.dumps(self._calls, indent=2)
+
+
+class TranscriptionLogger:
+    """Records final user transcriptions — written from a custom FrameProcessor."""
+
+    def __init__(self):
+        """Initialize the logger with an empty list of recorded transcriptions."""
+        self._entries: list[dict[str, Any]] = []
+
+    def log_transcription(self, text: str) -> None:
+        """Record a transcription.
+
+        Args:
+            text: The transcribed user utterance.
+        """
+        entry = {
+            "timestamp": datetime.now(UTC).isoformat(),
+            "text": text,
+        }
+        self._entries.append(entry)
+        logger.info(f"[TranscriptionLogger] {text!r}")
+
+    def dump(self) -> str:
+        """Return all recorded transcriptions as a JSON string."""
+        return json.dumps(self._entries, indent=2)
+
+
+@dataclass
+class AppResources:
+    """Typed container for everything the app shares across this session.
+
+    Add fields here as the app grows (e.g. ``db: AsyncConnection``,
+    ``http: httpx.AsyncClient``). Read sites ``cast()`` to this type to
+    get autocomplete and refactor safety:
+
+    - In tools: ``cast(AppResources, params.app_resources)``.
+    - In custom processors: ``cast(AppResources, self.pipeline_task.app_resources)``.
+    """
+
+    tool_call_logger: ToolCallLogger
+    transcription_logger: TranscriptionLogger
+
+
+async def fetch_weather_from_api(params: FunctionCallParams):
+    resources = cast(AppResources, params.app_resources)
+    resources.tool_call_logger.log_tool_call(params.function_name, params.arguments)
+    await params.result_callback({"conditions": "nice", "temperature": "75"})
+
+
+async def fetch_restaurant_recommendation(params: FunctionCallParams):
+    resources = cast(AppResources, params.app_resources)
+    resources.tool_call_logger.log_tool_call(params.function_name, params.arguments)
+    await params.result_callback({"name": "The Golden Dragon"})
+
+
+class TranscriptionLoggingProcessor(FrameProcessor):
+    """Logs each final user transcription into the shared app resources.
+
+    Demonstrates the second read site for ``app_resources``: any custom
+    ``FrameProcessor`` can reach the same bag every tool handler sees by
+    going through ``self.pipeline_task.app_resources``. ``pipeline_task``
+    is ``None`` until the task sets the processor up, so we guard against
+    that case.
+    """
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        """Forward all frames; log final user transcriptions on the way through."""
+        await super().process_frame(frame, direction)
+
+        if isinstance(frame, TranscriptionFrame) and self.pipeline_task is not None:
+            resources = cast(AppResources, self.pipeline_task.app_resources)
+            resources.transcription_logger.log_transcription(frame.text)
+
+        await self.push_frame(frame, direction)
+
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    stt = DeepgramSTTService(api_key=os.environ["DEEPGRAM_API_KEY"])
+
+    tts = CartesiaTTSService(
+        api_key=os.environ["CARTESIA_API_KEY"],
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
+    )
+
+    llm = OpenAIResponsesLLMService(
+        api_key=os.environ["OPENAI_API_KEY"],
+        settings=OpenAIResponsesLLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way.",
+        ),
+    )
+
+    # You can also register a function_name of None to get all functions
+    # sent to the same callback with an additional function_name parameter.
+    llm.register_function("get_current_weather", fetch_weather_from_api)
+    llm.register_function("get_restaurant_recommendation", fetch_restaurant_recommendation)
+
+    @llm.event_handler("on_connection_error")
+    async def on_connection_error(service, error):
+        logger.error(f"LLM connection error: {error}")
+
+    @llm.event_handler("on_function_calls_started")
+    async def on_function_calls_started(service, function_calls):
+        # Avoid appending this filler message to the LLM context — it would
+        # alter the conversation history and prevent
+        # OpenAIResponsesLLMService's previous_response_id optimization from
+        # matching, forcing a full context resend.
+        await tts.queue_frame(TTSSpeakFrame("Let me check on that.", append_to_context=False))
+
+    weather_function = FunctionSchema(
+        name="get_current_weather",
+        description="Get the current weather",
+        properties={
+            "location": {
+                "type": "string",
+                "description": "The city and state, e.g. San Francisco, CA",
+            },
+            "format": {
+                "type": "string",
+                "enum": ["celsius", "fahrenheit"],
+                "description": "The temperature unit to use. Infer this from the user's location.",
+            },
+        },
+        required=["location", "format"],
+    )
+    restaurant_function = FunctionSchema(
+        name="get_restaurant_recommendation",
+        description="Get a restaurant recommendation",
+        properties={
+            "location": {
+                "type": "string",
+                "description": "The city and state, e.g. San Francisco, CA",
+            },
+        },
+        required=["location"],
+    )
+    tools = ToolsSchema(standard_tools=[weather_function, restaurant_function])
+
+    context = LLMContext(tools=tools)
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),
+            stt,
+            TranscriptionLoggingProcessor(),
+            user_aggregator,
+            llm,
+            tts,
+            transport.output(),
+            assistant_aggregator,
+        ]
+    )
+
+    # Keep local handles so we can read collected state after the session
+    # ends; Pipecat never copies or clears the object.
+    tool_call_logger = ToolCallLogger()
+    transcription_logger = TranscriptionLogger()
+    resources = AppResources(
+        tool_call_logger=tool_call_logger,
+        transcription_logger=transcription_logger,
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+        app_resources=resources,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+        # Kick off the conversation.
+        context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+    # The session has ended; read whatever state the handlers built up.
+    logger.info(f"Tool calls logged during session:\n{tool_call_logger.dump()}")
+    logger.info(f"Transcriptions logged during session:\n{transcription_logger.dump()}")
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/features-before-and-after-events.py

@@ -1,5 +1,5 @@
 #
-# Copyright (c) 2024–2025, Daily
+# Copyright (c) 2024-2026, Daily
 #
 # SPDX-License-Identifier: BSD 2-Clause License
 #
@@ -10,16 +10,16 @@ from dataclasses import dataclass
 from dotenv import load_dotenv
 from loguru import logger
 
-from pipecat.audio.turn.smart_turn.base_smart_turn import SmartTurnParams
-from pipecat.audio.turn.smart_turn.local_smart_turn_v3 import LocalSmartTurnAnalyzerV3
 from pipecat.audio.vad.silero import SileroVADAnalyzer
-from pipecat.audio.vad.vad_analyzer import VADParams
 from pipecat.frames.frames import DataFrame, LLMRunFrame
 from pipecat.pipeline.pipeline import Pipeline
 from pipecat.pipeline.runner import PipelineRunner
 from pipecat.pipeline.task import PipelineParams, PipelineTask
 from pipecat.processors.aggregators.llm_context import LLMContext
-from pipecat.processors.aggregators.llm_response_universal import LLMContextAggregatorPair
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
 from pipecat.runner.types import RunnerArguments
 from pipecat.runner.utils import create_transport
 from pipecat.services.cartesia.tts import CartesiaTTSService
@@ -42,27 +42,20 @@ class CustomAfterPushFrame(DataFrame):
     pass
 
 
-# We store functions so objects (e.g. SileroVADAnalyzer) don't get
-# instantiated. The function will be called when the desired transport gets
-# selected.
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
 transport_params = {
     "daily": lambda: DailyParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "twilio": lambda: FastAPIWebsocketParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "webrtc": lambda: TransportParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
 }
 
@@ -70,34 +63,37 @@ transport_params = {
 async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     logger.info(f"Starting bot")
 
-    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+    stt = DeepgramSTTService(api_key=os.environ["DEEPGRAM_API_KEY"])
 
     tts = CartesiaTTSService(
-        api_key=os.getenv("CARTESIA_API_KEY"),
-        voice_id="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        api_key=os.environ["CARTESIA_API_KEY"],
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
     )
 
-    llm = OpenAILLMService(api_key=os.getenv("OPENAI_API_KEY"))
+    llm = OpenAILLMService(
+        api_key=os.environ["OPENAI_API_KEY"],
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way.",
+        ),
+    )
 
-    messages = [
-        {
-            "role": "system",
-            "content": "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be spoken aloud, so avoid special characters that can't easily be spoken, such as emojis or bullet points. Respond to what the user said in a creative and helpful way.",
-        },
-    ]
-
-    context = LLMContext(messages)
-    context_aggregator = LLMContextAggregatorPair(context)
+    context = LLMContext()
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
 
     pipeline = Pipeline(
         [
             transport.input(),  # Transport user input
             stt,
-            context_aggregator.user(),  # User responses
+            user_aggregator,  # User responses
             llm,  # LLM
             tts,  # TTS
             transport.output(),  # Transport bot output
-            context_aggregator.assistant(),  # Assistant spoken responses
+            assistant_aggregator,  # Assistant spoken responses
         ]
     )
 
@@ -124,7 +120,9 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        messages.append({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
         # Custom frames are pushed in order so they can be used for synchronization purposes.
         await task.queue_frames([CustomBeforeProcessFrame(), LLMRunFrame(), CustomAfterPushFrame()])
 

examples/features/features-concurrent-llm-evaluation.py

@@ -0,0 +1,173 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+
+import os
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import LLMRunFrame
+from pipecat.pipeline.parallel_pipeline import ParallelPipeline
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.processors.audio.vad_processor import VADProcessor
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.groq.llm import GroqLLMService
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+from pipecat.turns.user_turn_processor import UserTurnProcessor
+from pipecat.turns.user_turn_strategies import ExternalUserTurnStrategies
+
+load_dotenv(override=True)
+
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    stt = DeepgramSTTService(api_key=os.environ["DEEPGRAM_API_KEY"])
+
+    tts = CartesiaTTSService(
+        api_key=os.environ["CARTESIA_API_KEY"],
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
+    )
+
+    openai_llm = OpenAILLMService(
+        api_key=os.environ["OPENAI_API_KEY"],
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way.",
+        ),
+    )
+
+    groq_llm = GroqLLMService(
+        api_key=os.environ["GROQ_API_KEY"],
+        settings=GroqLLMService.Settings(
+            system_instruction="You are a very helpful assistant. Your goal is to demonstrate your capabilities in detail in a creative and helpful way.",
+        ),
+    )
+
+    openai_context = LLMContext()
+    groq_context = LLMContext()
+
+    # We use an external VADProcessor because the UserTurnProcessor is shared
+    # across multiple parallel aggregators. The VADProcessor emits
+    # VADUserStartedSpeakingFrame and VADUserStoppedSpeakingFrame which the
+    # UserTurnProcessor needs to manage turn lifecycle.
+    vad_processor = VADProcessor(vad_analyzer=SileroVADAnalyzer())
+
+    # We use this external user turn processor. This processor will push
+    # UserStartedSpeakingFrame and UserStoppedSpeakingFrame as well as
+    # interruptions. This can be used in advanced cases when there are multiple
+    # aggregators in the pipeline.
+    user_turn_processor = UserTurnProcessor()
+
+    # We use external user turn strategies for both aggregators since the turn
+    # management is done by the common UserTurnProcessor.
+    openai_context_aggregator = LLMContextAggregatorPair(
+        openai_context,
+        user_params=LLMUserAggregatorParams(user_turn_strategies=ExternalUserTurnStrategies()),
+    )
+    groq_context_aggregator = LLMContextAggregatorPair(
+        groq_context,
+        user_params=LLMUserAggregatorParams(user_turn_strategies=ExternalUserTurnStrategies()),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),  # Transport user input
+            stt,  # STT
+            vad_processor,
+            user_turn_processor,
+            ParallelPipeline(
+                [
+                    openai_context_aggregator.user(),  # User responses
+                    openai_llm,  # LLM
+                    tts,  # TTS (bot will speak the chosen language)
+                    transport.output(),  # Transport bot output
+                    openai_context_aggregator.assistant(),  # Assistant spoken responses
+                ],
+                [
+                    groq_context_aggregator.user(),  # User responses
+                    groq_llm,  # LLM
+                    groq_context_aggregator.assistant(),  # Assistant responses
+                ],
+            ),
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+        # Kick off the conversation.
+        openai_context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
+        groq_context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/features-concurrent-llm-rtvi-ignored-sources.py

@@ -0,0 +1,184 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+"""RTVIObserver ignored sources example.
+
+This example shows how to suppress RTVI messages from a specific pipeline
+processor so that secondary branches don't leak events to the client.
+
+"""
+
+import os
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import LLMRunFrame
+from pipecat.pipeline.parallel_pipeline import ParallelPipeline
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.processors.audio.vad_processor import VADProcessor
+from pipecat.processors.frameworks.rtvi import RTVIObserverParams
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+from pipecat.turns.user_turn_processor import UserTurnProcessor
+from pipecat.turns.user_turn_strategies import ExternalUserTurnStrategies
+
+load_dotenv(override=True)
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info("Starting bot")
+
+    stt = DeepgramSTTService(api_key=os.environ["DEEPGRAM_API_KEY"])
+
+    tts = CartesiaTTSService(
+        api_key=os.environ["CARTESIA_API_KEY"],
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
+    )
+
+    # Main LLM — drives the conversation. Its RTVI events reach the client.
+    main_llm = OpenAILLMService(
+        api_key=os.environ["OPENAI_API_KEY"],
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way.",
+        ),
+    )
+
+    # Evaluator LLM — silently grades the user's message in the background.
+    # Its RTVI events will be suppressed so the client is unaware of this branch.
+    evaluator_llm = OpenAILLMService(
+        api_key=os.environ["OPENAI_API_KEY"],
+        name="EvaluatorLLM",
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a silent quality evaluator. When given a user message, respond with a single JSON object: {'score': <1-5>, 'reason': '<brief reason>'}. Do not respond conversationally.",
+        ),
+    )
+
+    main_context = LLMContext()
+    evaluator_context = LLMContext()
+
+    # We use an external VADProcessor because the UserTurnProcessor is shared
+    # across multiple parallel aggregators. The VADProcessor emits
+    # VADUserStartedSpeakingFrame and VADUserStoppedSpeakingFrame which the
+    # UserTurnProcessor needs to manage turn lifecycle.
+    vad_processor = VADProcessor(vad_analyzer=SileroVADAnalyzer())
+
+    # We use this external user turn processor. This processor will push
+    # UserStartedSpeakingFrame and UserStoppedSpeakingFrame as well as
+    # interruptions. This can be used in advanced cases when there are multiple
+    # aggregators in the pipeline.
+    user_turn_processor = UserTurnProcessor()
+
+    # We use external user turn strategies for both aggregators since the turn
+    # management is done by the common UserTurnProcessor.
+    main_context_aggregator = LLMContextAggregatorPair(
+        main_context,
+        user_params=LLMUserAggregatorParams(user_turn_strategies=ExternalUserTurnStrategies()),
+    )
+    evaluator_context_aggregator = LLMContextAggregatorPair(
+        evaluator_context,
+        user_params=LLMUserAggregatorParams(user_turn_strategies=ExternalUserTurnStrategies()),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),  # Transport user input
+            stt,  # STT
+            vad_processor,
+            user_turn_processor,
+            ParallelPipeline(
+                # Main branch: speaks to the user.
+                [
+                    main_context_aggregator.user(),
+                    main_llm,
+                    tts,
+                    transport.output(),
+                    main_context_aggregator.assistant(),
+                ],
+                # Evaluator branch: silent background scoring, no audio output.
+                [
+                    evaluator_context_aggregator.user(),
+                    evaluator_llm,
+                    evaluator_context_aggregator.assistant(),
+                ],
+            ),
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        rtvi_observer_params=RTVIObserverParams(ignored_sources=[evaluator_llm]),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info("Client connected")
+        main_context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
+        evaluator_context.add_message(
+            {"role": "developer", "content": "Ready to evaluate user messages."}
+        )
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info("Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/features-custom-frame-processor.py

@@ -1,20 +1,15 @@
 #
-# Copyright (c) 2024–2025, Daily
+# Copyright (c) 2024-2026, Daily
 #
 # SPDX-License-Identifier: BSD 2-Clause License
 #
 
-import io
 import os
-import re
 
 from dotenv import load_dotenv
 from loguru import logger
 
-from pipecat.audio.turn.smart_turn.base_smart_turn import SmartTurnParams
-from pipecat.audio.turn.smart_turn.local_smart_turn_v3 import LocalSmartTurnAnalyzerV3
 from pipecat.audio.vad.silero import SileroVADAnalyzer
-from pipecat.audio.vad.vad_analyzer import VADParams
 from pipecat.frames.frames import (
     Frame,
     LLMRunFrame,
@@ -24,7 +19,10 @@ from pipecat.pipeline.pipeline import Pipeline
 from pipecat.pipeline.runner import PipelineRunner
 from pipecat.pipeline.task import PipelineParams, PipelineTask
 from pipecat.processors.aggregators.llm_context import LLMContext
-from pipecat.processors.aggregators.llm_response_universal import LLMContextAggregatorPair
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
 from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
 from pipecat.runner.types import RunnerArguments
 from pipecat.runner.utils import create_transport
@@ -75,22 +73,17 @@ class MetricsFrameLogger(FrameProcessor):
             await self.push_frame(frame, direction)
 
 
-# We store functions so objects (e.g. SileroVADAnalyzer) don't get
-# instantiated. The function will be called when the desired transport gets
-# selected.
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
 transport_params = {
     "daily": lambda: DailyParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "webrtc": lambda: TransportParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
         video_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
 }
 
@@ -98,24 +91,27 @@ transport_params = {
 async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     logger.info(f"Starting bot")
 
-    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+    stt = DeepgramSTTService(api_key=os.environ["DEEPGRAM_API_KEY"])
 
     tts = CartesiaTTSService(
-        api_key=os.getenv("CARTESIA_API_KEY"),
-        voice_id="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        api_key=os.environ["CARTESIA_API_KEY"],
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
     )
 
-    llm = OpenAILLMService(api_key=os.getenv("OPENAI_API_KEY"))
+    llm = OpenAILLMService(
+        api_key=os.environ["OPENAI_API_KEY"],
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way.",
+        ),
+    )
 
-    messages = [
-        {
-            "role": "system",
-            "content": "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be spoken aloud, so avoid special characters that can't easily be spoken, such as emojis or bullet points. Respond to what the user said in a creative and helpful way.",
-        },
-    ]
-
-    context = LLMContext(messages)
-    context_aggregator = LLMContextAggregatorPair(context)
+    context = LLMContext()
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
 
     metrics_frame_processor = MetricsFrameLogger()
 
@@ -123,11 +119,11 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         [
             transport.input(),
             stt,
-            context_aggregator.user(),
+            user_aggregator,
             llm,
             tts,
             transport.output(),
-            context_aggregator.assistant(),
+            assistant_aggregator,
             metrics_frame_processor,  # pretty print metrics frames
         ]
     )
@@ -145,7 +141,9 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected: {client}")
         # Kick off the conversation.
-        messages.append({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/features/features-gpu-container-local-bot.py

@@ -1,5 +1,5 @@
 #
-# Copyright (c) 2024–2025, Daily
+# Copyright (c) 2024-2026, Daily
 #
 # SPDX-License-Identifier: BSD 2-Clause License
 #
@@ -10,16 +10,16 @@ import os
 from dotenv import load_dotenv
 from loguru import logger
 
-from pipecat.audio.turn.smart_turn.base_smart_turn import SmartTurnParams
-from pipecat.audio.turn.smart_turn.local_smart_turn_v3 import LocalSmartTurnAnalyzerV3
 from pipecat.audio.vad.silero import SileroVADAnalyzer
-from pipecat.audio.vad.vad_analyzer import VADParams
 from pipecat.frames.frames import LLMRunFrame
 from pipecat.pipeline.pipeline import Pipeline
 from pipecat.pipeline.runner import PipelineRunner
 from pipecat.pipeline.task import PipelineParams, PipelineTask
 from pipecat.processors.aggregators.llm_context import LLMContext
-from pipecat.processors.aggregators.llm_response_universal import LLMContextAggregatorPair
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
 from pipecat.runner.types import RunnerArguments
 from pipecat.runner.utils import create_transport
 from pipecat.services.deepgram.stt import DeepgramSTTService
@@ -35,27 +35,20 @@ from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
 
 load_dotenv(override=True)
 
-# We store functions so objects (e.g. SileroVADAnalyzer) don't get
-# instantiated. The function will be called when the desired transport gets
-# selected.
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
 transport_params = {
     "daily": lambda: DailyParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "twilio": lambda: FastAPIWebsocketParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "webrtc": lambda: TransportParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
 }
 
@@ -63,41 +56,42 @@ transport_params = {
 async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     logger.info(f"Starting bot")
 
-    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+    stt = DeepgramSTTService(api_key=os.environ["DEEPGRAM_API_KEY"])
 
     tts = DeepgramTTSService(
-        api_key=os.getenv("DEEPGRAM_API_KEY"),
-        voice="aura-asteria-en",
+        api_key=os.environ["DEEPGRAM_API_KEY"],
+        settings=DeepgramTTSService.Settings(
+            voice="aura-asteria-en",
+        ),
         base_url="http://0.0.0.0:8080",
     )
 
     llm = OpenAILLMService(
         # To use OpenAI
-        # api_key=os.getenv("OPENAI_API_KEY"),
+        # api_key=os.environ["OPENAI_API_KEY"],
         # Or, to use a local vLLM (or similar) api server
-        model="meta-llama/Meta-Llama-3-8B-Instruct",
+        settings=OpenAILLMService.Settings(
+            model="meta-llama/Meta-Llama-3-8B-Instruct",
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way.",
+        ),
         base_url="http://0.0.0.0:8000/v1",
     )
 
-    messages = [
-        {
-            "role": "system",
-            "content": "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be spoken aloud, so avoid special characters that can't easily be spoken, such as emojis or bullet points. Respond to what the user said in a creative and helpful way.",
-        },
-    ]
-
-    context = LLMContext(messages)
-    context_aggregator = LLMContextAggregatorPair(context)
+    context = LLMContext()
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
 
     pipeline = Pipeline(
         [
             transport.input(),  # Transport user input
             stt,  # STT
-            context_aggregator.user(),
+            user_aggregator,
             llm,  # LLM
             tts,  # TTS
             transport.output(),  # Transport bot output
-            context_aggregator.assistant(),
+            assistant_aggregator,
         ]
     )
 
@@ -115,7 +109,9 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        messages.append({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
         await task.queue_frames([LLMRunFrame()])
 
     # Handle "latency-ping" messages. The client will send app messages that look like

examples/features/features-live-translation.py

@@ -1,25 +1,25 @@
 #
-# Copyright (c) 2024–2025, Daily
+# Copyright (c) 2024-2026, Daily
 #
 # SPDX-License-Identifier: BSD 2-Clause License
 #
 
+
 import os
 
 from dotenv import load_dotenv
 from loguru import logger
 
-from pipecat.audio.turn.smart_turn.base_smart_turn import SmartTurnParams
-from pipecat.audio.turn.smart_turn.local_smart_turn_v3 import LocalSmartTurnAnalyzerV3
 from pipecat.audio.vad.silero import SileroVADAnalyzer
-from pipecat.audio.vad.vad_analyzer import VADParams
 from pipecat.frames.frames import TTSSpeakFrame
 from pipecat.pipeline.pipeline import Pipeline
 from pipecat.pipeline.runner import PipelineRunner
 from pipecat.pipeline.task import PipelineParams, PipelineTask
 from pipecat.processors.aggregators.llm_context import LLMContext
-from pipecat.processors.aggregators.llm_response_universal import LLMContextAggregatorPair
-from pipecat.processors.filters.wake_check_filter import WakeCheckFilter
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
 from pipecat.runner.types import RunnerArguments
 from pipecat.runner.utils import create_transport
 from pipecat.services.cartesia.tts import CartesiaTTSService
@@ -28,30 +28,26 @@ from pipecat.services.openai.llm import OpenAILLMService
 from pipecat.transports.base_transport import BaseTransport, TransportParams
 from pipecat.transports.daily.transport import DailyParams
 from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+from pipecat.turns.user_start import TranscriptionUserTurnStartStrategy
+from pipecat.turns.user_turn_strategies import UserTurnStrategies
 
 load_dotenv(override=True)
 
-# We store functions so objects (e.g. SileroVADAnalyzer) don't get
-# instantiated. The function will be called when the desired transport gets
-# selected.
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
 transport_params = {
     "daily": lambda: DailyParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "twilio": lambda: FastAPIWebsocketParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "webrtc": lambda: TransportParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
 }
 
@@ -59,37 +55,47 @@ transport_params = {
 async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     logger.info(f"Starting bot")
 
-    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+    stt = DeepgramSTTService(api_key=os.environ["DEEPGRAM_API_KEY"])
 
     tts = CartesiaTTSService(
-        api_key=os.getenv("CARTESIA_API_KEY"),
-        voice_id="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        api_key=os.environ["CARTESIA_API_KEY"],
+        settings=CartesiaTTSService.Settings(
+            voice="d4db5fb9-f44b-4bd1-85fa-192e0f0d75f9",  # Spanish-speaking Lady
+        ),
     )
 
-    llm = OpenAILLMService(api_key=os.getenv("OPENAI_API_KEY"))
+    llm = OpenAILLMService(
+        api_key=os.environ["OPENAI_API_KEY"],
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a live translation assistant. Your sole purpose is to translate English text into Spanish. When you receive English text from the user, immediately translate it into natural, fluent Spanish. Do not add explanations, commentary, or extra information—only provide the Spanish translation of the text you receive.",
+        ),
+    )
 
-    messages = [
-        {
-            "role": "system",
-            "content": "You are a helpful assistant. Respond to what the user said in a creative and helpful way. Keep your responses brief.",
-        },
-    ]
+    context = LLMContext()
 
-    hey_robot_filter = WakeCheckFilter(["hey robot", "hey, robot"])
-
-    context = LLMContext(messages)
-    context_aggregator = LLMContextAggregatorPair(context)
+    # We use the TranscriptionUserTurnStartStrategy to start a new user turn
+    # every time a transcription is received. We disable interruptions, so the
+    # user can continue speaking while the bot is transcribing, without
+    # interrupting the bot.
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(
+            user_turn_strategies=UserTurnStrategies(
+                start=[TranscriptionUserTurnStartStrategy(enable_interruptions=False)],
+            ),
+            vad_analyzer=SileroVADAnalyzer(),
+        ),
+    )
 
     pipeline = Pipeline(
         [
             transport.input(),  # Transport user input
             stt,  # STT
-            hey_robot_filter,  # Filter out speech not directed at the robot
-            context_aggregator.user(),  # User responses
+            user_aggregator,  # User responses
             llm,  # LLM
-            tts,  # TTS
+            tts,  # TTS (bot will speak the chosen language)
             transport.output(),  # Transport bot output
-            context_aggregator.assistant(),  # Assistant spoken responses
+            assistant_aggregator,  # Assistant spoken responses
         ]
     )
 
@@ -105,8 +111,14 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     @transport.event_handler("on_client_connected")
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
-        # Kick off the conversation.
-        await task.queue_frame(TTSSpeakFrame("Hi! If you want to talk to me, just say 'Hey Robot'"))
+        await task.queue_frames(
+            [
+                TTSSpeakFrame(
+                    text="Hello, welcome to live translation. Everything you say will be automatically translated to Spanish. Let's begin!",
+                    append_to_context=True,
+                ),
+            ]
+        )
 
     @transport.event_handler("on_client_disconnected")
     async def on_client_disconnected(transport, client):

examples/features/features-pattern-pair-voice-switching.py

@@ -1,5 +1,5 @@
 #
-# Copyright (c) 2024–2025, Daily
+# Copyright (c) 2024-2026, Daily
 #
 # SPDX-License-Identifier: BSD 2-Clause License
 #
@@ -24,7 +24,7 @@ The PatternPairAggregator:
     - Returns processed text at sentence boundaries
 
 Requirements:
-    - OpenAI API key (for GPT-4o)
+    - OpenAI API key
     - Cartesia API key (for text-to-speech)
     - Daily API key (for video/audio transport)
 
@@ -44,16 +44,17 @@ import os
 from dotenv import load_dotenv
 from loguru import logger
 
-from pipecat.audio.turn.smart_turn.base_smart_turn import SmartTurnParams
-from pipecat.audio.turn.smart_turn.local_smart_turn_v3 import LocalSmartTurnAnalyzerV3
 from pipecat.audio.vad.silero import SileroVADAnalyzer
-from pipecat.audio.vad.vad_analyzer import VADParams
-from pipecat.frames.frames import LLMRunFrame
+from pipecat.frames.frames import LLMRunFrame, TTSUpdateSettingsFrame
 from pipecat.pipeline.pipeline import Pipeline
 from pipecat.pipeline.runner import PipelineRunner
 from pipecat.pipeline.task import PipelineParams, PipelineTask
 from pipecat.processors.aggregators.llm_context import LLMContext
-from pipecat.processors.aggregators.llm_response_universal import LLMContextAggregatorPair
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.processors.aggregators.llm_text_processor import LLMTextProcessor
 from pipecat.runner.types import RunnerArguments
 from pipecat.runner.utils import create_transport
 from pipecat.services.cartesia.tts import CartesiaTTSService
@@ -78,27 +79,20 @@ VOICE_IDS = {
     "male": "7cf0e2b1-8daf-4fe4-89ad-f6039398f359",  # Male character voice
 }
 
-# We store functions so objects (e.g. SileroVADAnalyzer) don't get
-# instantiated. The function will be called when the desired transport gets
-# selected.
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
 transport_params = {
     "daily": lambda: DailyParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "twilio": lambda: FastAPIWebsocketParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "webrtc": lambda: TransportParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
 }
 
@@ -107,42 +101,45 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     logger.info(f"Starting bot")
 
     # Create pattern pair aggregator for voice switching
-    pattern_aggregator = PatternPairAggregator()
+    llm_text_aggregator = PatternPairAggregator()
 
     # Add pattern for voice switching
-    pattern_aggregator.add_pattern(
+    llm_text_aggregator.add_pattern(
         type="voice",
         start_pattern="<voice>",
         end_pattern="</voice>",
-        action=MatchAction.REMOVE,  # Remove tags from final text
+        action=MatchAction.AGGREGATE,
     )
 
     # Register handler for voice switching
     async def on_voice_tag(match: PatternMatch):
         voice_name = match.text.strip().lower()
         if voice_name in VOICE_IDS:
-            # First flush any existing audio to finish the current context
-            await tts.flush_audio()
-            # Then set the new voice
-            tts.set_voice(VOICE_IDS[voice_name])
+            await llm_text_processor.push_frame(
+                TTSUpdateSettingsFrame(
+                    delta=CartesiaTTSService.Settings(voice=VOICE_IDS[voice_name])
+                )
+            )
             logger.info(f"Switched to {voice_name} voice")
         else:
             logger.warning(f"Unknown voice: {voice_name}")
 
-    pattern_aggregator.on_pattern_match("voice", on_voice_tag)
+    llm_text_aggregator.on_pattern_match("voice", on_voice_tag)
 
-    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+    stt = DeepgramSTTService(api_key=os.environ["DEEPGRAM_API_KEY"])
+
+    # Process LLM text through the pattern aggregator before TTS
+    llm_text_processor = LLMTextProcessor(text_aggregator=llm_text_aggregator)
 
     # Initialize TTS with narrator voice as default
     tts = CartesiaTTSService(
-        api_key=os.getenv("CARTESIA_API_KEY"),
-        voice_id=VOICE_IDS["narrator"],
-        text_aggregator=pattern_aggregator,
+        api_key=os.environ["CARTESIA_API_KEY"],
+        settings=CartesiaTTSService.Settings(
+            voice=VOICE_IDS["narrator"],
+        ),
+        skip_aggregator_types=["voice"],  # Skip voice tags in TTS speech
     )
 
-    # Initialize LLM
-    llm = OpenAILLMService(api_key=os.getenv("OPENAI_API_KEY"))
-
     # System prompt for storytelling with voice switching
     system_prompt = """You are an engaging storyteller that uses different voices to bring stories to life.
 
@@ -191,27 +188,31 @@ FOLLOW THESE RULES:
 
 Remember: Use narrator voice for EVERYTHING except the actual quoted dialogue."""
 
-    # Set up LLM context
-    messages = [
-        {
-            "role": "system",
-            "content": system_prompt,
-        },
-    ]
+    # Initialize LLM
+    llm = OpenAILLMService(
+        api_key=os.environ["OPENAI_API_KEY"],
+        settings=OpenAILLMService.Settings(
+            system_instruction=system_prompt,
+        ),
+    )
 
-    context = LLMContext(messages)
-    context_aggregator = LLMContextAggregatorPair(context)
+    context = LLMContext()
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
 
     # Create pipeline
     pipeline = Pipeline(
         [
             transport.input(),
             stt,
-            context_aggregator.user(),
+            user_aggregator,
             llm,
-            tts,  # TTS with pattern aggregator
+            llm_text_processor,
+            tts,
             transport.output(),
-            context_aggregator.assistant(),
+            assistant_aggregator,
         ]
     )
 

examples/features/features-service-switcher.py

@@ -1,5 +1,5 @@
 #
-# Copyright (c) 2024–2025, Daily
+# Copyright (c) 2024-2026, Daily
 #
 # SPDX-License-Identifier: BSD 2-Clause License
 #
@@ -12,18 +12,18 @@ from loguru import logger
 
 from pipecat.adapters.schemas.function_schema import FunctionSchema
 from pipecat.adapters.schemas.tools_schema import ToolsSchema
-from pipecat.audio.turn.smart_turn.base_smart_turn import SmartTurnParams
-from pipecat.audio.turn.smart_turn.local_smart_turn_v3 import LocalSmartTurnAnalyzerV3
 from pipecat.audio.vad.silero import SileroVADAnalyzer
-from pipecat.audio.vad.vad_analyzer import VADParams
 from pipecat.frames.frames import LLMRunFrame, ManuallySwitchServiceFrame
 from pipecat.pipeline.llm_switcher import LLMSwitcher
 from pipecat.pipeline.pipeline import Pipeline
 from pipecat.pipeline.runner import PipelineRunner
-from pipecat.pipeline.service_switcher import ServiceSwitcher, ServiceSwitcherStrategyManual
+from pipecat.pipeline.service_switcher import ServiceSwitcher
 from pipecat.pipeline.task import PipelineParams, PipelineTask
 from pipecat.processors.aggregators.llm_context import LLMContext
-from pipecat.processors.aggregators.llm_response_universal import LLMContextAggregatorPair
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
 from pipecat.runner.types import RunnerArguments
 from pipecat.runner.utils import create_transport
 from pipecat.services.cartesia.stt import CartesiaSTTService
@@ -56,27 +56,20 @@ async def get_restaurant_recommendation(params: FunctionCallParams, location: st
     await params.result_callback({"name": "The Golden Dragon"})
 
 
-# We store functions so objects (e.g. SileroVADAnalyzer) don't get
-# instantiated. The function will be called when the desired transport gets
-# selected.
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
 transport_params = {
     "daily": lambda: DailyParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "twilio": lambda: FastAPIWebsocketParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "webrtc": lambda: TransportParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
 }
 
@@ -101,51 +94,60 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         required=["location", "format"],
     )
 
-    stt_cartesia = CartesiaSTTService(api_key=os.getenv("CARTESIA_API_KEY"))
-    stt_deepgram = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
-    stt_switcher = ServiceSwitcher(
-        services=[stt_cartesia, stt_deepgram], strategy_type=ServiceSwitcherStrategyManual
-    )
+    stt_cartesia = CartesiaSTTService(api_key=os.environ["CARTESIA_API_KEY"])
+    stt_deepgram = DeepgramSTTService(api_key=os.environ["DEEPGRAM_API_KEY"])
+    # Uses ServiceSwitcherStrategyManual by default
+    stt_switcher = ServiceSwitcher(services=[stt_cartesia, stt_deepgram])
 
     tts_cartesia = CartesiaTTSService(
-        api_key=os.getenv("CARTESIA_API_KEY"),
-        voice_id="71a7ad14-091c-4e8e-a314-022ece01c121",
+        api_key=os.environ["CARTESIA_API_KEY"],
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
     )
-    tts_deepgram = DeepgramTTSService(api_key=os.getenv("DEEPGRAM_API_KEY"))
-    tts_switcher = ServiceSwitcher(
-        services=[tts_cartesia, tts_deepgram], strategy_type=ServiceSwitcherStrategyManual
+    tts_deepgram = DeepgramTTSService(
+        api_key=os.environ["DEEPGRAM_API_KEY"],
+        settings=DeepgramTTSService.Settings(
+            voice="aura-2-helena-en",
+        ),
     )
+    # Uses ServiceSwitcherStrategyManual by default
+    tts_switcher = ServiceSwitcher(services=[tts_cartesia, tts_deepgram])
 
-    llm_openai = OpenAILLMService(api_key=os.getenv("OPENAI_API_KEY"))
-    llm_google = GoogleLLMService(api_key=os.getenv("GOOGLE_API_KEY"))
-    llm_switcher = LLMSwitcher(
-        llms=[llm_openai, llm_google], strategy_type=ServiceSwitcherStrategyManual
+    system_prompt = "You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way."
+
+    llm_openai = OpenAILLMService(
+        api_key=os.environ["OPENAI_API_KEY"],
+        settings=OpenAILLMService.Settings(system_instruction=system_prompt),
     )
+    llm_google = GoogleLLMService(
+        api_key=os.environ["GOOGLE_API_KEY"],
+        settings=GoogleLLMService.Settings(system_instruction=system_prompt),
+    )
+    # Uses ServiceSwitcherStrategyManual by default
+    llm_switcher = LLMSwitcher(llms=[llm_openai, llm_google])
     # Register a "classic" function
     llm_switcher.register_function("get_current_weather", fetch_weather_from_api)
     # Register a "direct" function
     llm_switcher.register_direct_function(get_restaurant_recommendation)
 
-    messages = [
-        {
-            "role": "system",
-            "content": "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be spoken aloud, so avoid special characters that can't easily be spoken, such as emojis or bullet points. Respond to what the user said in a creative and helpful way.",
-        },
-    ]
     tools = ToolsSchema(standard_tools=[weather_function, get_restaurant_recommendation])
 
-    context = LLMContext(messages, tools)
-    context_aggregator = LLMContextAggregatorPair(context)
+    context = LLMContext(tools=tools)
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
 
     pipeline = Pipeline(
         [
             transport.input(),  # Transport user input
             stt_switcher,
-            context_aggregator.user(),  # User responses
+            user_aggregator,  # User responses
             llm_switcher,  # LLM
             tts_switcher,  # TTS
             transport.output(),  # Transport bot output
-            context_aggregator.assistant(),  # Assistant spoken responses
+            assistant_aggregator,  # Assistant spoken responses
         ]
     )
 
@@ -162,7 +164,9 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        messages.append({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
         await task.queue_frames([LLMRunFrame()])
         await asyncio.sleep(15)
         print(f"Switching to {stt_deepgram}")

examples/features/features-switch-languages.py

@@ -1,5 +1,5 @@
 #
-# Copyright (c) 2024–2025, Daily
+# Copyright (c) 2024-2026, Daily
 #
 # SPDX-License-Identifier: BSD 2-Clause License
 #
@@ -7,23 +7,22 @@
 
 import os
 
-from deepgram import LiveOptions
 from dotenv import load_dotenv
 from loguru import logger
 
 from pipecat.adapters.schemas.function_schema import FunctionSchema
 from pipecat.adapters.schemas.tools_schema import ToolsSchema
-from pipecat.audio.turn.smart_turn.base_smart_turn import SmartTurnParams
-from pipecat.audio.turn.smart_turn.local_smart_turn_v3 import LocalSmartTurnAnalyzerV3
 from pipecat.audio.vad.silero import SileroVADAnalyzer
-from pipecat.audio.vad.vad_analyzer import VADParams
 from pipecat.frames.frames import Frame, LLMRunFrame
 from pipecat.pipeline.parallel_pipeline import ParallelPipeline
 from pipecat.pipeline.pipeline import Pipeline
 from pipecat.pipeline.runner import PipelineRunner
 from pipecat.pipeline.task import PipelineParams, PipelineTask
 from pipecat.processors.aggregators.llm_context import LLMContext
-from pipecat.processors.aggregators.llm_response_universal import LLMContextAggregatorPair
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
 from pipecat.processors.filters.function_filter import FunctionFilter
 from pipecat.runner.types import RunnerArguments
 from pipecat.runner.utils import create_transport
@@ -43,13 +42,17 @@ class SwitchLanguage(ParallelPipeline):
         self._current_language = "English"
 
         english_tts = CartesiaTTSService(
-            api_key=os.getenv("CARTESIA_API_KEY"),
-            voice_id="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+            api_key=os.environ["CARTESIA_API_KEY"],
+            settings=CartesiaTTSService.Settings(
+                voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+            ),
         )
 
         spanish_tts = CartesiaTTSService(
-            api_key=os.getenv("CARTESIA_API_KEY"),
-            voice_id="d4db5fb9-f44b-4bd1-85fa-192e0f0d75f9",  # Spanish-speaking Lady
+            api_key=os.environ["CARTESIA_API_KEY"],
+            settings=CartesiaTTSService.Settings(
+                voice="d4db5fb9-f44b-4bd1-85fa-192e0f0d75f9",  # Spanish-speaking Lady
+            ),
         )
 
         super().__init__(
@@ -76,27 +79,20 @@ class SwitchLanguage(ParallelPipeline):
         return self.current_language == "Spanish"
 
 
-# We store functions so objects (e.g. SileroVADAnalyzer) don't get
-# instantiated. The function will be called when the desired transport gets
-# selected.
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
 transport_params = {
     "daily": lambda: DailyParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "twilio": lambda: FastAPIWebsocketParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "webrtc": lambda: TransportParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
 }
 
@@ -105,12 +101,20 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     logger.info(f"Starting bot")
 
     stt = DeepgramSTTService(
-        api_key=os.getenv("DEEPGRAM_API_KEY"), live_options=LiveOptions(language="multi")
+        api_key=os.environ["DEEPGRAM_API_KEY"],
+        settings=DeepgramSTTService.Settings(
+            language="multi",
+        ),
     )
 
     tts = SwitchLanguage()
 
-    llm = OpenAILLMService(api_key=os.getenv("OPENAI_API_KEY"))
+    llm = OpenAILLMService(
+        api_key=os.environ["OPENAI_API_KEY"],
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way. You can speak the following languages: 'English' and 'Spanish'.",
+        ),
+    )
     llm.register_function("switch_language", tts.switch_language)
 
     switch_language_function = FunctionSchema(
@@ -125,25 +129,21 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         required=["language"],
     )
     tools = ToolsSchema(standard_tools=[switch_language_function])
-    messages = [
-        {
-            "role": "system",
-            "content": "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities. Respond to what the user said in a creative and helpful way. Your output should not include non-alphanumeric characters. You can speak the following languages: 'English' and 'Spanish'.",
-        },
-    ]
-
-    context = LLMContext(messages, tools)
-    context_aggregator = LLMContextAggregatorPair(context)
+    context = LLMContext(tools=tools)
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
 
     pipeline = Pipeline(
         [
             transport.input(),  # Transport user input
             stt,  # STT
-            context_aggregator.user(),  # User responses
+            user_aggregator,  # User responses
             llm,  # LLM
             tts,  # TTS (bot will speak the chosen language)
             transport.output(),  # Transport bot output
-            context_aggregator.assistant(),  # Assistant spoken responses
+            assistant_aggregator,  # Assistant spoken responses
         ]
     )
 
@@ -160,9 +160,9 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        messages.append(
+        context.add_message(
             {
-                "role": "system",
+                "role": "developer",
                 "content": f"Please introduce yourself to the user and let them know the languages you speak. Your initial responses should be in {tts.current_language}.",
             }
         )

examples/features/features-switch-voices.py

@@ -1,5 +1,5 @@
 #
-# Copyright (c) 2024–2025, Daily
+# Copyright (c) 2024-2026, Daily
 #
 # SPDX-License-Identifier: BSD 2-Clause License
 #
@@ -12,17 +12,17 @@ from loguru import logger
 
 from pipecat.adapters.schemas.function_schema import FunctionSchema
 from pipecat.adapters.schemas.tools_schema import ToolsSchema
-from pipecat.audio.turn.smart_turn.base_smart_turn import SmartTurnParams
-from pipecat.audio.turn.smart_turn.local_smart_turn_v3 import LocalSmartTurnAnalyzerV3
 from pipecat.audio.vad.silero import SileroVADAnalyzer
-from pipecat.audio.vad.vad_analyzer import VADParams
 from pipecat.frames.frames import Frame, LLMRunFrame
 from pipecat.pipeline.parallel_pipeline import ParallelPipeline
 from pipecat.pipeline.pipeline import Pipeline
 from pipecat.pipeline.runner import PipelineRunner
 from pipecat.pipeline.task import PipelineParams, PipelineTask
 from pipecat.processors.aggregators.llm_context import LLMContext
-from pipecat.processors.aggregators.llm_response_universal import LLMContextAggregatorPair
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
 from pipecat.processors.filters.function_filter import FunctionFilter
 from pipecat.runner.types import RunnerArguments
 from pipecat.runner.utils import create_transport
@@ -42,18 +42,24 @@ class SwitchVoices(ParallelPipeline):
         self._current_voice = "News Lady"
 
         news_lady = CartesiaTTSService(
-            api_key=os.getenv("CARTESIA_API_KEY"),
-            voice_id="bf991597-6c13-47e4-8411-91ec2de5c466",  # Newslady
+            api_key=os.environ["CARTESIA_API_KEY"],
+            settings=CartesiaTTSService.Settings(
+                voice="bf991597-6c13-47e4-8411-91ec2de5c466",  # Newslady
+            ),
         )
 
         british_lady = CartesiaTTSService(
-            api_key=os.getenv("CARTESIA_API_KEY"),
-            voice_id="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+            api_key=os.environ["CARTESIA_API_KEY"],
+            settings=CartesiaTTSService.Settings(
+                voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+            ),
         )
 
         barbershop_man = CartesiaTTSService(
-            api_key=os.getenv("CARTESIA_API_KEY"),
-            voice_id="a0e99841-438c-4a64-b679-ae501e7d6091",  # Barbershop Man
+            api_key=os.environ["CARTESIA_API_KEY"],
+            settings=CartesiaTTSService.Settings(
+                voice="a0e99841-438c-4a64-b679-ae501e7d6091",  # Barbershop Man
+            ),
         )
 
         super().__init__(
@@ -87,27 +93,20 @@ class SwitchVoices(ParallelPipeline):
         return self.current_voice == "Barbershop Man"
 
 
-# We store functions so objects (e.g. SileroVADAnalyzer) don't get
-# instantiated. The function will be called when the desired transport gets
-# selected.
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
 transport_params = {
     "daily": lambda: DailyParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "twilio": lambda: FastAPIWebsocketParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "webrtc": lambda: TransportParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
-        vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
 }
 
@@ -115,11 +114,16 @@ transport_params = {
 async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     logger.info(f"Starting bot")
 
-    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+    stt = DeepgramSTTService(api_key=os.environ["DEEPGRAM_API_KEY"])
 
     tts = SwitchVoices()
 
-    llm = OpenAILLMService(api_key=os.getenv("OPENAI_API_KEY"))
+    llm = OpenAILLMService(
+        api_key=os.environ["OPENAI_API_KEY"],
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative and helpful way. You can do the following voices: 'News Lady', 'British Lady' and 'Barbershop Man'.",
+        ),
+    )
     llm.register_function("switch_voice", tts.switch_voice)
 
     switch_voice_function = FunctionSchema(
@@ -135,25 +139,21 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     )
     tools = ToolsSchema(standard_tools=[switch_voice_function])
 
-    messages = [
-        {
-            "role": "system",
-            "content": "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities. Respond to what the user said in a creative and helpful way. Your output should not include non-alphanumeric characters. You can do the following voices: 'News Lady', 'British Lady' and 'Barbershop Man'.",
-        },
-    ]
-
-    context = LLMContext(messages, tools)
-    context_aggregator = LLMContextAggregatorPair(context)
+    context = LLMContext(tools=tools)
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
 
     pipeline = Pipeline(
         [
             transport.input(),  # Transport user input
             stt,
-            context_aggregator.user(),  # User responses
+            user_aggregator,  # User responses
             llm,  # LLM
             tts,  # TTS with switch voice functionality
             transport.output(),  # Transport bot output
-            context_aggregator.assistant(),  # Assistant spoken responses
+            assistant_aggregator,  # Assistant spoken responses
         ]
     )
 
@@ -170,9 +170,9 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        messages.append(
+        context.add_message(
             {
-                "role": "system",
+                "role": "developer",
                 "content": f"Please introduce yourself to the user and let them know the voices you can do. Your initial responses should be as if you were a {tts.current_voice}.",
             }
         )