Installation¶
Requirements¶
| Requirement | Version | Purpose |
|---|---|---|
| Python | 3.12+ | Runtime |
| Docker | 24+ | Sandbox execution, local infrastructure |
| Dapr CLI | 1.14+ | Agent sidecar |
Install Grampus¶
Optional extras¶
# Anthropic Claude support
pip install "grampus-ai[anthropic]"
# OpenAI GPT support
pip install "grampus-ai[openai]"
# Both providers
pip install "grampus-ai[all]"
Verify the install¶
Install the Dapr CLI¶
Dapr provides the infrastructure runtime (state, pub/sub, workflows). Grampus requires the Dapr sidecar to be running alongside your agent.
Initialize Dapr (installs the runtime components locally):
Verify:
Start local infrastructure¶
Grampus requires PostgreSQL (with pgvector) and Redis. Use the provided docker-compose.yml:
# From your project directory (generated by grampus init)
docker compose up -d
# Verify all services are healthy
docker compose ps
Expected output:
Environment variables¶
Set your LLM provider API key:
# Anthropic
export GRAMPUS_MODEL__ANTHROPIC_API_KEY="sk-ant-..."
# OpenAI
export GRAMPUS_MODEL__OPENAI_API_KEY="sk-..."
Using a .env file
Grampus reads .env files automatically. Create .env in your project root and add your keys there — they will not be committed to git if .env is in .gitignore.
Next steps¶
- Quickstart → — Build and run your first agent in 5 minutes
- Configuration reference → — Full
GrampusConfigfield reference