bash-mcp
Bash MCP
Documentation
bash-mcp
A simple MCP (Model Context Protocol) server that enables Claude to execute shell commands without permission prompts.
⚠️ Security Warning: This server executes arbitrary shell commands. Use with caution and only in trusted environments.
Installation
# Install globally
npm install -g bash-mcp
# Or use with npx
npx bash-mcpQuick Start
For Claude Desktop
Add to your `claude_desktop_config.json`:
{
"mcpServers": {
"bash": {
"command": "npx",
"args": ["bash-mcp"]
}
}
}For Claude Code (Cursor, VS Code)
1. Open command palette (Cmd/Ctrl + Shift + P)
2. Run "MCP: Add Server"
3. Select "NPM" as the server type
4. Enter: `bash-mcp`
Available Tools
`run` - Execute a command
// Simple command
run("ls -la")
// With working directory
run("npm test", { cwd: "/path/to/project" })
// With timeout (milliseconds)
run("long-running-command", { timeout: 60000 })`run_background` - Start a background process
// Start a dev server
run_background("npm run dev", "frontend")
// Start backend service with working directory
run_background("./gradlew bootRun", "backend", { cwd: "./backend" })`kill_background` - Stop a background process
kill_background("frontend")`list_background` - List all background processes
list_background()Example Usage
User: Start the development servers
Assistant: I'll start both frontend and backend servers for you.
[Uses run_background tool]
Started frontend server (PID: 12345)
Started backend server (PID: 12346)
User: Check if they're running
Assistant: [Uses list_background tool]
Both servers are running successfully!Response Format
All tools return JSON formatted responses:
{
"success": true,
"stdout": "command output",
"stderr": "error output if any",
"command": "executed command"
}For background processes:
{
"success": true,
"name": "frontend",
"pid": 12345,
"command": "npm run dev",
"message": "Started background process 'frontend' (PID: 12345)"
}Features
- Execute any shell command without permission prompts
- Run long-running processes in the background
- Manage background processes (list, kill)
- Capture stdout and stderr
- Set working directory for commands
- Configure timeout for commands
- Automatic cleanup on server shutdown
- NEW: Automatic output truncation with full output saved to temp files
- NEW: Configurable output size limits and temp directory via environment variables
Environment Variables
- `BASH_MCP_MAX_OUTPUT_SIZE`: Maximum output size in bytes before truncation (default: 51200/50KB)
- `BASH_MCP_TEMP_DIR`: Directory for storing full output when truncated (default: system temp directory)
Example Configuration
{
"mcpServers": {
"bash": {
"command": "npx",
"args": ["bash-mcp"],
"env": {
"BASH_MCP_MAX_OUTPUT_SIZE": "102400",
"BASH_MCP_TEMP_DIR": "/tmp/bash-mcp-outputs"
}
}
}
}Output Overflow Handling
When command output exceeds `BASH_MCP_MAX_OUTPUT_SIZE`:
1. Output is truncated to the specified limit
2. Full output is saved to a temporary file
3. The response includes the file path where full output can be found
4. If custom temp directory fails, falls back to system temp directory
Security Considerations
This MCP server executes arbitrary shell commands with the same privileges as the Node.js process. Only use in development environments or trusted contexts.
Requirements
- Node.js >= 16.0.0
- npm or npx
License
MIT
Author
tinywind
Contributing
Issues and pull requests are welcome at GitHub.
Frequently asked questions
What is bash-mcp?
bash-mcp is Bash MCP
How do I install bash-mcp?
Open the GitHub repository and follow its README. Most MCP servers are added to your client's MCP config, then called by your agent.
Is bash-mcp open source?
Yes — it is hosted on GitHub at https://github.com/tinywind/bash-mcp and has 11 stars.
Related MCP tools
🔥 Official Firecrawl MCP Server - Adds powerful web scraping and search to Cursor, Claude and any other LLM clients.
Use any LLMs (Large Language Models) for Deep Research. Support SSE API and MCP server.
Enhanced MCP server for interactive user feedback and command execution in AI-assisted development, featuring dual interface support (Web UI and Desktop Application) with intelligent environment detection and cross-platform compatibility.
A powerful Zotero AI and MCP plugin with ChatGPT, Gemini 3.7, Claude Fable 5, Claude Opus 5, DeepSeek V4, Grok, OpenRouter, Kimi k3, GLM 5.3, SiliconFlow, GPT-oss, Gemma 4, Qwen 3.8
Connect your browser to AI models. Just use Dia on Chrome, Arc or Firefox.
文颜 MCP Server 可以让 AI 自动将 Markdown 文章排版后发布至微信公众号。
Run your own MCP server? See who uses it and what to fix.
Measure it with TrackMCP