Complyra = Comply + RA(G) — Enterprise compliance knowledge assistant powered by Retrieval-Augmented Generation.

Complyra combines multi-tenant retrieval-augmented generation, human-in-the-loop approval workflow, RBAC, full audit logging, and cloud-ready deployment automation — built for compliance-sensitive environments.

graph LR
    subgraph Frontend
        UI[React SPA]
    end

    subgraph API Layer
        GW[FastAPI Gateway]
        Auth[JWT + RBAC]
    end

    subgraph Workflow
        LG[LangGraph]
        R[Retrieve]
        D[Draft]
        P[Policy Gate]
        A[Approval]
    end

    subgraph Data
        QD[Qdrant Vector DB]
        PG[PostgreSQL]
        RD[Redis Queue]
    end

    subgraph LLM
        OL[Ollama]
        EMB[Embeddings BGE/OpenAI]
    end

    subgraph Observability
        PR[Prometheus]
        GR[Grafana]
        LS[LangSmith]
        SE[Sentry]
    end

    UI -->|HTTP/SSE| GW
    GW --> Auth --> LG
    LG --> R --> QD
    LG --> D --> OL
    LG --> P
    LG --> A --> PG
    R --> EMB
    GW -->|Audit| PG
    GW -->|Ingest Jobs| RD
    GW -->|/metrics| PR
    PR --> GR
    LG -.->|Traces| LS
    GW -.->|Errors| SE

• Multi-tenant RAG — tenant-scoped document ingestion and retrieval via X-Tenant-ID
• Human-in-the-loop approval — LangGraph workflow with configurable approval gates
• Pluggable embeddings — SentenceTransformer (BGE) or OpenAI API, switchable via config
• SSE streaming — real-time token-by-token chat via POST /chat/stream
• RBAC — three roles: admin, auditor, user
• Output policy guard — regex-based detection of secrets, API keys, and credentials
• Audit trail — full event logging with search and CSV export
• Async ingestion — Redis + RQ worker for background document processing
• LangSmith tracing — optional LLM observability with zero code overhead
• Observability — Prometheus metrics, Grafana dashboards, Sentry error tracking

git clone https://github.com/complyra/complyra.git
cd complyra
cp .env.example .env
docker compose up --build -d

Default credentials: demo / demo123

# Backend
python3 -m venv .venv && source .venv/bin/activate
pip install -r requirements-dev.txt
cp .env.example .env
alembic upgrade head
uvicorn app.main:app --host 0.0.0.0 --port 8000 --reload

# Frontend
cd web && npm install && npm run dev

All settings use the APP_ prefix. See .env.example for the full list.

See docs/streaming-api.md for the SSE streaming protocol.

pip install -r requirements-dev.txt
PYTHONPATH=. pytest tests/ -v --cov=app --cov-report=term-missing

black --check app/
isort --check app/
ruff check app/

1. Create and secure AWS account (docs/aws-account-onboarding.md)
2. Prepare production env (./scripts/aws/00_preflight.sh)
3. Terraform plan + policy gate (./scripts/aws/07_terraform_plan.sh)
4. Build/push images (./scripts/aws/03_build_and_push.sh)
5. Deploy services (./scripts/aws/09_deploy_services_from_release.sh)
6. Run smoke tests (./scripts/aws/05_smoke_test.sh)

Full runbook: docs/aws-deployment.md

See CONTRIBUTING.md.

See SECURITY.md.

MIT