Ragdoll

Tech Stack

• 🚀 Backend: Python (FastAPI), Redis, PostgreSQL
• 💻 Frontend: TypeScript, Vite + React
• 🛠️ Infrastructure: Docker, DBmate

Ragdoll is a self-hosted knowledge workspace. It ingests documents, builds searchable and graph-backed context, keeps evidence and corrections attached to answers, and exposes that state through search, chat, entities, pinned facts, changes, and admin surfaces.

License: Business Source License 1.1
Free for personal use and internal organizational use; see Commercial Terms
Apache License 2.0 on 2030-01-01

Architecture Visual

flowchart TD
  Upload["Document upload"] --> Ingestion["Ingestion API"]
  Ingestion --> Queue["Queue and worker"]
  Queue --> Processing["Text, embedding, and entity processing"]
  Processing --> Storage["Postgres + pgvector"]
  Processing --> Graph["Graph projection"]
  Storage --> Search["Search"]
  Search --> Chat["Chat"]
  Graph --> Entities["Entities"]

Loading

This is the fast, high-level story: documents move through ingestion and background processing into retrieval and graph projections that power the product experience. For the canonical detailed flow, start with docs/architecture/ingestion-and-processing.md and then use the broader architecture guide.

Product Overview

The current application includes:

• a FastAPI backend in apps/api
• a Vite + React frontend in apps/web
• shared OpenAPI and TypeScript contracts in packages/contracts
• DBmate-managed relational schema under apps/api/db
• Redis-backed background processing for document ingestion
• self-hosted runtime dependencies wired through infra/docker

Primary capability areas:

• auth, users, and spaces
• document library, upload, processing, and download
• search, entities, and knowledge graph exploration
• chat with citations and correction submission
• pinned facts and change tracking
• admin runtime visibility and usage reporting

Repository Structure

ragdoll-redux/
  apps/
    api/
    web/
  packages/
    contracts/
    config/
    tooling/
  tests/
    e2e/
  infra/
    docker/
    supabase/
    ollama/
  scripts/
    dev/
    test/
    ops/
  docs/

Read INDEX.md before broad codebase traversal. It is the placement guide for pages, modules, tests, contracts, and scripts.

Documentation Guide

Start with the current-state docs:

• Architecture: docs/architecture/README.md
• Executive overview: docs/executive/system-brief.md
• Operating model and risk: docs/executive/operating-model-and-risk.md
• Engineering subsystem docs:
  • docs/engineering/identity-and-access.md
  • docs/engineering/documents-and-ingestion.md
  • docs/engineering/retrieval-and-knowledge.md
  • docs/engineering/interaction-and-governance.md
  • docs/engineering/api-surface-and-module-ownership.md

Local Development

Use ./dev-setup.sh as the main entrypoint.

• Start infra: ./dev-setup.sh infra up
• Start the application stack: ./dev-setup.sh daemon
• Inspect services: ./dev-setup.sh ps
• Run the full repo test suite: ./dev-setup.sh test
• Run end-to-end coverage: ./dev-setup.sh test-e2e
• Stop the app stack: ./dev-setup.sh down
• Stop infra: ./dev-setup.sh infra down

Startup helpers create these files when missing:

• apps/api/.env from apps/api/.env.example
• apps/web/.env from apps/web/.env.example
• infra/docker/.env.infra from infra/docker/.env.infra.example

Testing expectations live in TESTING.md.

Runtime Shape

The logged-out surface stays intentionally small:

• / and /login
• /register
• /status

Authenticated users move through dashboard, spaces, documents, search, chat, entities, pinned-facts, changes, and account. Admin users also get /admin.

Docs Maintenance

Canonical current-state docs live in:

• docs/architecture/ for system architecture
• docs/executive/ for high-level leadership-facing docs
• docs/engineering/ for codebase-facing subsystem docs

Use Mermaid diagrams in those areas whenever a flow, dependency, or ownership boundary is easier to understand visually than in prose.