Garcon
Garcon is a self-hosted browser and mobile workspace that enables users to run and steer multiple parallel AI coding agent sessions simultaneously. It supports various popular coding assistants including Claude Code, Codex, Cursor Agent, OpenCode, Amp, Droid, and Pi, providing a unified interface to manage their outputs and interactions. The platform includes integrated tools such as a terminal, file browser, diff review, and Git/PR integration, allowing developers to seamlessly review and integrate AI-generated code changes directly from the workspace. By orchestrating multiple agents, Garcon facilitates complex multi-step coding tasks that can be distributed across different agents for parallel processing. With its self-hosted nature, Garcon offers privacy and control over the environment, making it suitable for teams that need to keep code and workflows on their own infrastructure. The workspace is accessible via web and mobile, enabling flexible monitoring and steering of agent sessions from anywhere.
- Category
- AI Agents
- Pricing
- Open Source
Problem solved
Coordinating multiple AI coding assistants for complex software projects can be chaotic without a central management interface. Garcon solves this by providing a unified, self-hosted workspace where users can run, monitor, and steer several AI agent sessions in parallel, with integrated version control tools.
Use cases
- Running multiple AI agents to simultaneously refactor different modules of a codebase.
- Reviewing AI-generated code diffs before merging changes.
- Coordinating parallel bug-fixing sessions across different parts of a project.
- Steering AI agents via a unified terminal and file system while they generate code.
- Managing Git workflows and pull requests directly from the agent workspace.
- Using mobile access to monitor and adjust agent sessions on the go.
Audience
- Developers
- ML Engineers
- Researchers
Pros
- Centralized orchestration of multiple AI coding agents.
- Seamless integration with Git and pull requests.
- Self-hosted ensuring data privacy and control.
- Web and mobile accessibility for flexible management.
- Supports a wide range of popular coding agents.
Cons
- Requires self-hosting setup, which may be complex for non-technical users.
- Potential performance overhead when running multiple parallel sessions.
- Limited to agents that are compatible with the workspace.
- Documentation may be sparse for initial configuration.
- No cloud-hosted option, which might deter users seeking convenience.
Official website: https://github.com/cfal/garcon
Canonical LambdaBase page: https://www.lambdabase.com/tools/garcon