claude-code-mcp Project MCP Server

@KunihiroS

リポジトリホームページ

基本情報

カテゴリ

AI とエージェント

ライセンス

MIT

トランスポート

stdio

作者

KunihiroS

投稿者

Kunihiro Sugiyama

claude-code-mcp Project MCP Server について

MCP Server connects with claude code local command.

#mcp #claude #anthropic #anthropic-claude #model-context-protocol #mcp-server #claude-code

claude-code-mcp Project MCP Server の設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

{
  "mcpServers": {
    "claude-code-mcp": {
      "command": "npx",
      "args": [
        "@kunihiros/claude-code-mcp"
      ]
    }
  }
}

claude-code-mcp Project MCP Server のツール

Provides a detailed explanation of the given code.

Reviews the given code.

Fixes bugs or issues in the given code.

Edits the given code based on instructions.

Generates tests for the given code.

Simulates the execution of a given command.

Sends a custom query with context.

概要

What is claude-code-mcp?

claude-code-mcp is an MCP server that calls the locally installed Claude Code CLI and provides tools for code explanation, review, fixing, editing, testing, command simulation, and custom queries. It communicates via stdio and uses Base64 encoding to handle special characters in natural language text.

How to use claude-code-mcp?

Run the server via npx @kunihiros/claude-code-mcp or install globally with npm install -g claude-code-mcp and run claude-code-mcp. Set the required CLAUDE_BIN environment variable to the path of your Claude CLI executable. Configuration can be provided through MCP host settings, a .env file, or a global ~/.claude-code-mcp.env file.

Key features of claude-code-mcp

Provides seven tools: explain_code, review_code, fix_code, edit_code, test_code, simulate_command, and your_own_query
Base64 encodes natural language input for stable processing
Communicates via stdio using JSON‑formatted requests and responses
Configurable log level (debug, info, warn, error)
Supports running from MCP hosts like Claude Desktop (unconfirmed)

Use cases of claude-code-mcp

Explain or review code using Claude Code from any MCP‑compatible host
Automatically fix bugs or edit code with natural language instructions
Generate unit tests for existing code
Simulate command execution before running it
Send custom queries with optional context

よくある質問

What are the runtime requirements?

Node.js 18 or later (tested with v22.14.0), npm (or yarn), and the Claude Code CLI installed and authenticated.

How do I set the path to the Claude CLI?

Set the `CLAUDE_BIN` environment variable (required) to the absolute path of your Claude executable, e.g. `/home/linuxbrew/.linuxbrew/bin/claude`.

What tools are available?

The server offers `explain_code`, `review_code`, `fix_code`, `edit_code`, `test_code`, `simulate_command`, and `your_own_query`.

Can I use claude-code-mcp with Claude Desktop?

The README notes that calling Claude Code from Claude Desktop may work, but this is unconfirmed.

Where are logs stored?

The log file `claude-code-mcp.log` is first created in the project root, then falls back to the user’s home directory (`~/.claude-code-mcp.log`), and finally to `/tmp/claude-code-mcp.log`. Log rotation is not implemented.

「AI とエージェント」の他のコンテンツ

CrawlForge MCP Server

mysleekdesigns

28-tool MCP server for web scraping, crawling, deep research & autonomous extraction — clean Markdown & structured JSON for Claude, Cursor & any MCP client. 1,000 free one-time credits, local-Ollama LLM support.

q-ring

I4cTime

Quantum-inspired keyring for AI coding agents. Secure secrets with superposition, entanglement, tunneling, and teleportation — CLI + 44 MCP tools.

Hostinger

hostinger

Model Context Protocol (MCP) server for Hostinger API. If you don't want to install or run anything locally, connect directly to Hostinger's hosted MCP server: This opens a browser window to authorize via OAuth. Once app

148

Legion MCP

faulkj

MCP-native LLM councils for debates, juries, blind panels, voting, refinement, and custom multi-model deliberation.

strac-mcp-dlp

strac-io

Detect and redact PII, PHI, PCI and secrets — SSNs, credit cards, passports, API keys and cloud credentials — in text and in local files, using the Strac DLP API. Five tools: redact_text, detect_sensitive_data, detect_fi