cherry-mcp

What This Skill Does

Cherry MCP acts as a vital HTTP bridge designed specifically for the OpenClaw ecosystem. Many modern AI tools are built using the Model Context Protocol (MCP), which traditionally relies on stdio streams. These processes are inherently fragile; if the client process disconnects, the MCP server terminates immediately. Cherry MCP solves this by spawning and managing these servers as persistent child processes, ensuring they remain alive regardless of the client's state. It then wraps these server capabilities into a standardized REST API, allowing OpenClaw agents to interact with any MCP-compliant toolset via simple HTTP requests.

Installation

To integrate this skill into your environment, execute the following command in your terminal: clawhub install openclaw/skills/skills/bitbrujo/cherry-mcp

Once installed, you configure servers via the CLI: node cli.js add-server <name> <command> [args...]. It is highly recommended to manage the bridge process using pm2 to ensure the bridge itself is supervised and auto-restarts upon system reboots.

Use Cases

Cherry MCP is perfect for developers who want to bring external capabilities into their OpenClaw agents without writing custom middleware. Common use cases include integrating GitHub repositories, managing local databases, or accessing third-party APIs that have existing MCP implementations. It allows an agent to maintain a consistent 'workspace' of tools that are ready for immediate execution.

Example Prompts

1. "Cherry MCP, list all the tools currently available in the github server configuration so I can see what functions I have access to."
2. "Use the github tool to search for open pull requests in the openclaw/core repository using the query: 'authentication fix'."
3. "Restart the mcp-filesystem server through the bridge to clear the current connection state and refresh the tool definitions."

Tips & Limitations

Security is paramount: the bridge defaults to 127.0.0.1 binding and supports rate limiting and audit logging. Never expose the bridge to a public network directly. Be mindful that environment variables stored via set-env reside in config.json in plain text; for sensitive API keys, inject them directly via your shell environment before launching the bridge to keep your credentials secure. Finally, remember that you are in control: the bridge only executes pre-configured commands defined in your config file, preventing unauthorized arbitrary command execution.