mcp_shell_tools
A full-featured MCP server for local development with filesystem, shell, editor, session persistence, and security features.
README
mcp_shell_tools
A full-featured MCP server for local development — filesystem, shell, editor, session persistence.
Built as a practical companion to the blog series on uc-it.de about building MCP servers from scratch. This is not a demo — it runs in daily production use with Claude Desktop, claude.ai, and ChatGPT.
What Makes This Different
Most MCP servers offer a handful of tools over stdio. This one provides:
- 24 tools across 7 categories (filesystem, editor, search, shell, project context, memory, sessions)
- Two transports — stdio for Claude Desktop, Streamable HTTP for remote clients
- Session persistence — memory and context survive across conversations
- Security layer — dangerous command patterns are blocked, sudo requires confirmation
- Process isolation — subprocesses run in separate process groups with timeout enforcement
- OAuth 2.0 — optional authentication for remote access via Traefik reverse proxy
Tools
| Category | Tools |
|---|---|
| Filesystem | file_read, file_write, file_list, glob_search, file_delete, file_move, file_copy, file_append, tree |
| Editor | str_replace, diff_preview |
| Search | grep |
| Shell | shell_exec, env, set_env |
| Project | cd, cwd, project_init |
| Memory | memory_add, memory_show, memory_clear |
| Session | session_save, session_resume, session_list |
Quickstart
git clone https://github.com/cuber-it/mcp_shell_tools.git
cd mcp_shell_tools
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
Claude Desktop (stdio)
Add to your Claude Desktop config (~/.config/claude/claude_desktop_config.json):
{
"mcpServers": {
"shell-tools": {
"command": "/path/to/mcp_shell_tools/run.sh"
}
}
}
Restart Claude Desktop. The tools are now available.
Streamable HTTP (Remote)
python code/main.py serve --http 12201
This starts the server on port 12201 with the /mcp endpoint. For production use with OAuth, see the systemd service template in mcp-shell-http.service.
How It Works
Claude Desktop ──► run.sh ──► mcp_shell_tools (stdio)
claude.ai / ChatGPT
│
▼
Reverse Proxy ──► host:12201 (/mcp)
│
OAuth Server ──► host:9080
The server manages a working directory, environment variables, and session state. Tools operate relative to the current working directory (changeable via cd). Project context is loaded from CLAUDE.md files. Memory entries persist across tool calls within a session.
Security
Shell commands pass through a security filter before execution:
- Blocked patterns —
rm -rf /,dd of=/dev/,mkfs, fork bombs, etc. - Sudo warning — requires explicit confirmation
- Process groups — each subprocess runs in its own session, enabling clean kills on timeout
- Output truncation — prevents context overflow from large outputs
See code/config.py for the full list of blocked patterns.
Configuration
| Variable | Default | Description |
|---|---|---|
MCP_HOST |
127.0.0.1 |
Bind address (HTTP mode) |
MCP_PORT |
12201 |
Port (HTTP mode) |
MCP_OAUTH_ENABLED |
false |
Enable OAuth authentication |
MCP_OAUTH_SERVER_URL |
— | OAuth server URL |
MCP_PUBLIC_URL |
— | Public-facing server URL |
Project Structure
mcp_shell_tools/
├── code/
│ ├── main.py # CLI entry point (stdio / HTTP)
│ ├── server.py # FastMCP setup, tool registration
│ ├── config.py # Constants, security patterns
│ ├── state.py # Working directory, env vars
│ ├── health_server.py # HTTP health endpoint
│ ├── heinzel_integration.py # Optional registry integration
│ ├── tools/
│ │ ├── filesystem.py # 9 file operations
│ │ ├── editor.py # str_replace, diff_preview
│ │ ├── search.py # grep with regex support
│ │ ├── shell.py # Command execution with security
│ │ ├── project.py # Working directory, CLAUDE.md
│ │ ├── memory.py # Persistent notes and decisions
│ │ ├── session.py # Save/resume sessions
│ │ └── commands.py # Slash commands (/verbose, /log)
│ ├── persistence/
│ │ ├── models.py # Data models (Session, Memory)
│ │ └── session_manager.py # JSON-based persistence
│ └── utils/
├── tests/ # 70 tests
├── config/
│ └── traefik-mcp.yml # Example Traefik config
├── docs/
│ ├── ARCHITECTURE.md
│ ├── HTTP_STABILITY.md
│ └── ROADMAP.md
└── mcp-shell-http.service # Systemd service template
Requirements
- Python 3.10+
mcpSDK >= 1.26.0
Further Reading
- Blog: Building MCP Servers — the blog series this project accompanies
- Model Context Protocol — MCP specification
docs/ARCHITECTURE.md— detailed code architecturedocs/HTTP_STABILITY.md— transport evolution and Traefik setupCHANGELOG.md— version history
License
MIT
Author
UC IT Service — uc-it.de
Actively maintained and continuously evolving. Contributions and feedback welcome.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.