Forra REST API

The Forra backend exposes a REST API covering chat completion, assistants, conversations, skills, memories, vision, and more.

API Reference

The full API reference is auto-generated from the backend's pydantic models — every endpoint's request and response schemas reflect exactly what the server accepts and returns.

• Browse the API reference — endpoints grouped by resource
• Shared schemas — types referenced across endpoints
• OpenAPI 3.1 specification (openapi.json) — machine-readable spec for SDK generation, Insomnia/Bruno import, etc.

How the docs are generated

Each backend blueprint has its endpoints declared with @spec.validate(...) decorators that attach pydantic models for the request and response. A generator script dumps the OpenAPI spec and renders one markdown page per blueprint. The output is regenerated on every build.

Key capabilities

• Assistants and conversations (CRUD, streaming chat completion)
• Skills (global and assistant-scoped, upload, build-async)
• Memories (user + assistant-scoped)
• Vision (multimodal chat with image uploads)
• Embeddings and search
• Admin audit logs, marketplace templates, notifications

Authentication

Most endpoints require a JWT access token. See the /api/auth section (coming soon in the generated docs) for details.