claude-mcp

Claude MCP Server

claude-mcp connects AI tools to Claude Code for programmatic queries and conversation follow-ups. It serves developers building agent workflows that need synchronous replies or asynchronous tasks with status checks and cancellation.

Server facts

claude-mcp

Project

Author
ebeloded
License
MIT

Tech

Runtime / language
node
Transports
stdio
Remote available
No
Authentication
Not specified

Activity

GitHub stars
13
Last commit
Data as of

Overview

Claude Code conversations can become part of another AI tool's workflow through claude-mcp. Its listed tools are query, which exposes control over SDK options, status for inspecting a task by ID, and cancel for stopping an active task. The README describes both synchronous requests and asynchronous execution, allowing callers to obtain a direct response or track work separately. Task monitoring includes elapsed time and results when execution finishes.

Conversation handling in the README centers on starting fresh work and resuming from a selected response ID. A fresh conversation can target a particular working directory and add text to the system prompt; follow-ups retain the original directory. Selecting an earlier response provides a branching point rather than relying on whichever conversation ran most recently. The project uses Node.js and communicates over stdio, with a configured Claude Code CLI as a prerequisite. Its documented permission behavior bypasses Claude Code permission prompts. Optional debug logging, macOS task notifications, and cleanup of orphaned processes on disconnection support operation and troubleshooting.

claude-mcp setup notes

  • claude-mcp records stdio only; confirm a network transport upstream before planning remote access.

Related MCP servers