Skip to content

Latest commit

 

History

History
116 lines (88 loc) · 3.7 KB

API_DOCUMENTATION.md

File metadata and controls

116 lines (88 loc) · 3.7 KB

API Documentation

Current backend API reference for Live Conversational Threads.

  • Base URL: http://localhost:8000
  • Interactive docs:
    • Swagger: http://localhost:8000/docs
    • ReDoc: http://localhost:8000/redoc

Source of truth is the mounted FastAPI routers in lct_python_backend/backend.py.

Auth and Security

  • HTTP auth is optional and environment-driven:
    • If AUTH_TOKEN is unset, requests are unauthenticated in local dev.
    • If AUTH_TOKEN is set, non-health endpoints require Authorization: Bearer <token>.
  • WebSocket auth uses ?token=<AUTH_TOKEN> when token auth is enabled.
  • Rate limiting and body-size limits are enabled in middleware and configurable via env vars.

WebSocket Endpoints

WS /ws/transcripts

Primary realtime endpoint for audio-to-transcript-to-graph flow.

Typical message sequence:

  1. Client connects to /ws/transcripts.
  2. Client sends session_meta.
  3. Client streams audio_chunk payloads.
  4. Server emits transcript and graph updates:
    • transcript_partial
    • transcript_final
    • existing_json
    • chunk_dict
    • processing_status
  5. Client sends final_flush before close.

GET /ws/audio (legacy)

Deprecated compatibility endpoint; returns HTTP 410 with guidance to use /ws/transcripts.

Core REST Endpoints

Import

  • GET /api/import/health
  • POST /api/import/google-meet
  • POST /api/import/google-meet/preview
  • POST /api/import/from-url (env-gated by ENABLE_URL_IMPORT)
  • POST /api/import/from-text

Conversations and Persistence

  • GET /conversations/
  • GET /conversations/{conversation_id}
  • DELETE /conversations/{conversation_id}
  • POST /save_json/
  • POST /get_chunks/
  • POST /generate-context-stream/
  • POST /generate_formalism/

STT + LLM Settings

  • GET /api/settings/stt
  • PUT /api/settings/stt
  • GET /api/settings/stt/telemetry
  • POST /api/settings/stt/health-check
  • GET /api/settings/llm
  • PUT /api/settings/llm
  • GET /api/settings/llm/models

Graph + Thematic

  • GET /api/graph/health
  • POST /api/graph/generate
  • GET /api/graph/{conversation_id}
  • GET /api/graph/{conversation_id}/nodes
  • GET /api/graph/{conversation_id}/edges
  • DELETE /api/graph/{conversation_id}
  • POST /api/conversations/{conversation_id}/themes/generate
  • GET /api/conversations/{conversation_id}/themes
  • GET /api/conversations/{conversation_id}/themes/levels

Analysis

  • POST /api/conversations/{conversation_id}/simulacra/analyze
  • GET /api/conversations/{conversation_id}/simulacra
  • POST /api/conversations/{conversation_id}/biases/analyze
  • GET /api/conversations/{conversation_id}/biases
  • POST /api/conversations/{conversation_id}/frames/analyze
  • GET /api/conversations/{conversation_id}/frames

Analytics and Bookmarks

  • GET /api/analytics/conversations/{conversation_id}/analytics
  • GET /api/analytics/conversations/{conversation_id}/timeline
  • GET /api/analytics/conversations/{conversation_id}/roles
  • POST /api/bookmarks
  • GET /api/bookmarks
  • PATCH /api/bookmarks/{bookmark_id}
  • DELETE /api/bookmarks/{bookmark_id}

Prompt Management

  • GET /api/prompts
  • GET /api/prompts/config
  • GET /api/prompts/{prompt_name}
  • PUT /api/prompts/{prompt_name}
  • POST /api/prompts/{prompt_name}/validate
  • POST /api/prompts/reload

Notes

  • Some routes return legacy payload shapes for compatibility with existing frontend flows.
  • POST /save_json/ currently writes through the GCS helper path; in local environments without configured ADC/bucket, this endpoint can return a 500 until cloud credentials are configured.
  • For latest request/response schemas, use Swagger (/docs) because those are generated from current Pydantic models and router signatures.