sidekick-cli

Your agentic CLI developer.

Sidekick is an agentic CLI-based AI tool inspired by Claude Code, Copilot, Windsurf and Cursor. It's meant to be an open source alternative to these tools, providing a similar experience but with the flexibility of using different LLM providers (Anthropic, OpenAI, Google Gemini) while keeping the agentic workflow.

Sidekick is currently in beta and under active development. Please report issues or share feedback!

• No vendor lock-in. Use whichever LLM provider you prefer.
• MCP support
• Easily switch between models in the same session.
• JIT-style system prompt injection ensures Sidekick doesn't lose the plot.
• Per-project guide. Adjust Sidekick's behavior to suit your needs.
• CLI-first design. Ditch the clunky IDE.
• Cost and token tracking.
• Per command or per session confirmation skipping.

• Tests 😅
• More LLM providers, including Ollama

Install Sidekick.

pip install sidekick-cli

On first run, you'll be asked to configure your LLM providers.

sidekick

After initial setup, Sidekick saves a config file to ~/.config/sidekick.json. You can open and edit this file as needed. Future updates will make editing easier directly from within Sidekick.

Sidekick supports Model Context Protocol (MCP) servers. You can configure MCP servers in your ~/.config/sidekick.json file:

{
  "mcpServers": {
    "fetch": {
      "command": "uvx",
      "args": ["mcp-server-fetch"]
    },
    "github": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-github"],
      "env": {
        "GITHUB_PERSONAL_ACCESS_TOKEN": "<YOUR_TOKEN>"
      }
    }
  }
}

MCP servers extend the capabilities of your AI assistant, allowing it to interact with additional tools and data sources. Learn more about MCP at modelcontextprotocol.io.

• /help - Show available commands
• /yolo - Toggle "yolo" mode (skip tool confirmations)
• /clear - Clear message history
• /model - List available models
• /model <num> - Switch to a specific model (by index)
• /usage - Show session usage statistics
• exit - Exit the application

Sidekick supports the use of a "guide". This is a SIDEKICK.md file in the project root that contains instructions for Sidekick. Helpful for specifying tech stack, project structure, development preferences etc.

Sidekick uses Sentry for error tracking and usage analytics. You can disable this by starting with the --no-telemetry flag.

sidekick --no-telemetry

• Python 3.10 or higher
• Git (for undo functionality)

pip install sidekick-cli

1. Clone the repository
2. Install dependencies: pip install . (or pip install -e . for development)

# Install development dependencies
make install

# Run linting
make lint

# Run tests
make test

When preparing a new release:

1. Update version numbers in:

   • pyproject.toml
   • src/sidekick/constants.py (APP_VERSION)

2. Commit the version changes:

   git add pyproject.toml src/sidekick/constants.py
   git commit -m "chore: bump version to X.Y.Z"

3. Create and push a tag:

   git tag vX.Y.Z
   git push origin vX.Y.Z

4. Create a GitHub release:

   gh release create vX.Y.Z --title "vX.Y.Z" --notes "Release notes here"

5. Merge to main branch and push to trigger PyPI release (automated)

This project follows the Conventional Commits specification for commit messages:

• feat: - New features
• fix: - Bug fixes
• docs: - Documentation changes
• style: - Code style changes (formatting, etc.)
• refactor: - Code refactoring
• perf: - Performance improvements
• test: - Test additions or modifications
• chore: - Maintenance tasks (version bumps, etc.)
• build: - Build system changes
• ci: - CI configuration changes

• PyPI Package
• GitHub Issues
• GitHub Repository

MIT