CrewAI Project Template

A Docker-first, multi-agent AI framework template using CrewAI, LiteLLM, Vault, Weaviate, and Redis.

Quick Start

1. Clone and rename:

   git clone <this-repo> my-new-project
   cd my-new-project
   # Replace "myproject" / "MyProject" with your project name throughout:
   grep -rl "myproject" . --include="*.py" --include="*.yml" --include="*.yaml" --include="*.sh" --include="*.md" --include="*.env*" | xargs sed -i 's/myproject/yourproject/g'
   grep -rl "MyProject" . --include="*.py" --include="*.md" | xargs sed -i 's/MyProject/YourProject/g'

2. Configure environment:

   cp .env.example .env
   # Edit .env and add your LLM provider API keys

3. Start services:

   chmod +x scripts/start.sh
   ./scripts/start.sh

4. Run agents:

   docker compose exec crewai python -m crews

What's Included

Directory	Purpose
agents/	Agent role definitions and LLM factory (get_llm)
tasks/	Reusable task templates
crews/	Crew assembly factories and interactive CLI
tools/	Custom CrewAI tools with Pydantic schemas
flows/	Multi-step flow orchestration
config/	LiteLLM proxy configuration
scripts/	Startup and health-check scripts
tests/	Pytest fixtures for testing without Docker

Docker Services

• LiteLLM (port 4000) — LLM proxy with model routing and fallbacks
• Vault (port 8200) — Secret management (dev mode)
• Redis (port 6379) — Agent communication and caching
• Weaviate (port 8080) — Vector store for RAG
• PostgreSQL (port 5432) — LiteLLM spend tracking
• t2v-transformers — Embedding model (MiniLM-L6-v2)

Testing

pip install -r requirements-dev.txt
pytest

Tests run without Docker. See tests/conftest.py for shared fixtures.

Documentation

See CLAUDE.md for full conventions on adding agents, crews, tasks, tools, and flows.