Skip to main content
All endpoints on this page require an admin role. Regular members will receive 403 Forbidden.

System configuration

Runtime configuration overrides are stored in Postgres (app_config table) and take precedence over config.yml. Changes take effect immediately — no restart needed (except switching back to local transcription on a Scaleway-booted worker).

Transcription config

GET /api/admin/system/transcription-config PUT /api/admin/system/transcription-config
client is local or scaleway. Scaleway credentials are ai_inference.scaleway and are not exposed here — set them in config.yml.

Rag config (hirag-config route)

GET /api/admin/system/hirag-config PUT /api/admin/system/hirag-config Controls the AI agent client, embedding client, LLM model, embedding model, and (when using Ollama) the Ollama base URL and optional API key. Changing the embedding model or dimension requires re-embedding all stored topic vectors. The route keeps its hirag-config name after the hirag → rag rename.

Summarisation config

GET /api/admin/system/summarisation-config PUT /api/admin/system/summarisation-config Controls the summarisation worker’s Ollama URL and optional API key (separate from the rag Ollama settings).

Jobs

List transcription jobs

GET /api/admin/jobs/transcriptions Paginated list of all transcription jobs across all workspaces. Shows:
  • Final jobs — one per session
  • Chunk groups — a rolled-up row per session showing chunk count, aggregate status, and mean progress
Chunk groups are non-clickable; they give an at-a-glance view of chunk processing alongside their parent final job. Statuses: pending, processing, completed, failed Failed chunks can be re-run individually from the admin UI.

Unified job list

GET /api/admin/jobs lists jobs of every type, filterable by job_type, status and a search term. It includes topic_graph jobs (queued after a session summary completes), document_graph jobs (queued after a document summary completes or a document is deleted; they cannot be re-run from the admin queue) and pattern_synthesis jobs, all run by the rag worker.

Session topic graph

GET /api/admin/sessions/{session_id}/topic-graph shows a session’s topics as the topic graph holds them. Graph health per workspace (topics, entities, themes, sessions, vector index status) and a vector search over topic embeddings are at GET /api/admin/system/graph-health and POST /api/admin/system/graph-search.

Queue depths

GET /api/admin/system/queues returns ready-message depths per queue, including topic_graph_queue and document_graph_queue.

User management

List users

GET /api/admin/users

Update user role

PUT /api/admin/users/{user_id}/role

Waitlist management

GET /api/admin/waitlist POST /api/admin/waitlist/{entry_id}/approve The waitlist holds sign-up requests that haven’t been approved yet. Approving an entry creates the user account and sends an invite.
organisation_id adds them to a team organisation as well; null means Personal only. A personal organisation’s id returns 400. Their Personal organisation and home workspace are created at their first login (User lifecycle).

Sessions (admin view)

Admins can see all sessions across all workspaces. GET /api/admin/sessions — full session list with workspace context This is separate from the member-scoped GET /api/sessions endpoint which only returns sessions for the authenticated user’s workspaces.