create-agent 🤖

One-command bootstrap for OpenAI Agents SDK projects.

Usage

# npx (no install needed)
npx create-agent my-project

# With agent type
npx create-agent my-project --type single   # default — single agent + tools
npx create-agent my-project --type multi    # multi-agent with handoffs
npx create-agent my-project --type voice    # voice pipeline (STT → agent → TTS)

# uvx (if you prefer Python toolchain)
uvx create-agent my-project

What you get

my-project/
├── main.py                          # Entry point — just run this
├── agents/                          # Agent definitions
├── tools/                           # @function_tool functions
├── tests/                           # Pytest tests (ready to run)
├── pyproject.toml                   # uv-managed deps
├── Makefile                         # make run / make test / make lint
├── .env.example                     # Copy → .env, add OPENAI_API_KEY
├── .gitignore
└── .github/copilot-instructions.md  # Copilot context — team onboarding

After bootstrap

cd my-project
cp .env.example .env    # add your OPENAI_API_KEY
uv sync                 # install deps
uv run main.py          # run your agent

Agent types

| Type | Description | |------|-------------| | single | Single agent with function tools. Best starting point. | | multi | Orchestrator + specialist agents connected via handoffs. | | voice | Voice pipeline skeleton (STT → agent → TTS). |

Copilot integration

Every project includes .github/copilot-instructions.md. When your teammate opens the project in VS Code with GitHub Copilot, it automatically loads the agent context — they can ask Copilot things like:

• "Add a tool that fetches weather data"
• "Add a guardrail that blocks off-topic questions"
• "Write a test for the orchestrator"
• "Add memory to persist conversation history"

Stack

• openai-agents — OpenAI Agents SDK
• uv — Python package manager
• pytest — testing
• ruff — linting + formatting

Author

Eddie Chan — github.com/unrealandychan