Move workflow evaluations to standalone runner

evals/README.md

@@ -0,0 +1,48 @@
+# Workflow evaluations
+
+The evaluation runner calls a live FastAPI service backed by FastGPT. For now,
+each conversation step checks only the SSE `stage_code.nextStageCode` value.
+
+## Case format
+
+Add YAML files anywhere under `evals/cases/`:
+
+```yaml
+name: browser_addon_example
+clientMode: browser_addon
+
+steps:
+  - input: hi
+    expect:
+      stageCode: browser_addon.1001
+
+  - input: 继续
+    expect:
+      stageCode: browser_addon.1002
+```
+
+Steps in one file share the same generated `sessionId`. Different cases always
+use different session IDs.
+
+## Run
+
+Start the API, then execute:
+
+```bash
+python evals/run.py
+```
+
+Use another API address:
+
+```bash
+python evals/run.py --base-url http://192.168.1.10:8000
+```
+
+Run selected cases:
+
+```bash
+python evals/run.py --case browser_addon
+```
+
+The process exits with code `1` when an expectation fails and code `2` when
+case configuration is invalid.