Skip to main content

AtlasML API

The AtlasML API is a versioned REST interface exposed via FastAPI. Base paths depend on deployment configuration (e.g., service root when accessed internally). All responses use standard HTTP status codes and Pydantic-validated payloads.

All endpoints are versioned and grouped by router.

Health​

  • GET /api/v1/health/
    • Returns 200 OK only when AtlasML can still reach Weaviate, the required collections exist, and a lightweight read succeeds.
    • Returns 503 Service Unavailable with component details when the service is up but its vector-database dependency is not usable.

Competency​

Authentication: if enabled upstream, include Authorization: <API_KEY> with a token present in ATLAS_API_KEYS. Unauthorized requests will be rejected by the dependency if configured on the route.

  • POST /api/v1/competency/suggest

    • Request: SuggestCompetencyRequest
    • Response: SuggestCompetencyResponse
    • Suggests relevant competencies for a given textual description and course context by leveraging the service’s ML pipeline and vector search. Use this to provide recommendations during content creation or tagging.

  • POST /api/v1/competency/save

    • Request: SaveCompetencyRequest
    • Response: 200 OK (no content)
    • Persists competencies and/or exercises with an explicit operation_type (UPDATE/DELETE). Use this to upsert items or remove outdated entries, triggering downstream updates where applicable.

  • GET /api/v1/competency/relations/suggest/{course_id}

    • Response: CompetencyRelationSuggestionResponse
    • Generates candidate directed relations between competencies for the specified course. This can support graph-building tasks such as prerequisite mapping or curriculum analysis.

See the models page for full schema definitions and field descriptions.