Installation

Python SDK

pip install tappass

Requires Python 3.11+. Check with python3 --version.

Optional extras:

pip install tappass[sandbox]       # + kernel-level sandbox (nono)
pip install tappass[pii]           # Presidio PII detection
pip install tappass[llm]           # LiteLLM multi-provider routing

Install everything at once:

pip install tappass[all]

Start the server

The server is included in the same package:

tappass up

This launches an interactive setup wizard that:

1. Asks for your LLM API key (OpenAI, Anthropic, or any OpenAI-compatible provider)
2. Creates a default governance pipeline
3. Starts the server on http://localhost:9620

Verify

tappass status

You should see:

TapPass v0.5.0
Status: running
Port:   9620
Health: http://localhost:9620/health/ready

Or check via HTTP:

curl http://localhost:9620/health/ready
# → {"status": "ok"}

Stop the server

tappass down

Or press Ctrl+C if running in the foreground.

Docker (alternative)

If you prefer Docker:

docker run -p 9620:9620 \
  -e TAPPASS_JWT_SECRET=$(openssl rand -base64 48) \
  -e TAPPASS_ADMIN_API_KEY=tp_admin_$(openssl rand -hex 16) \
  -e OPENAI_API_KEY=sk-... \
  ghcr.io/tappass/tappass:latest

For the full production stack (TapPass + OPA + PostgreSQL + Redis), see the Architecture page.

Troubleshooting

Problem	Fix
pip install tappass fails	Make sure you have Python 3.11+. Try pip3 install tappass
tappass up says “port in use”	Another process is using port 9620. Run lsof -i :9620 to find it
connection refused on :9620	Server isn’t running. Run tappass up first
OPENAI_API_KEY not set	The setup wizard will prompt you, or set it: export OPENAI_API_KEY=sk-...
Health endpoint returns 503	Server is starting up. Wait a few seconds and retry