Hatch is a CLI tool for configuring MCP servers across AI host platforms. Adding a server to Claude Desktop, Cursor, VS Code, and others normally means editing separate JSON config files in different locations. Hatch does it from one command.
It also has a package system for managing MCP servers with dependency isolation, though that part is still being developed — see Getting Started for the current state.
Current status: suitable for development and trusted environments. Not hardened for production or multi-tenant use yet — see Security and Trust.
- Configure MCP servers on one or more AI host platforms at once
- Discover which host platforms are installed on your machine
- List and inspect server registrations across all your tools
- Manage MCP server packages with dependency isolation (system, Python, Docker)
Claude Desktop, Claude Code, VS Code, Cursor, Kiro, Codex, LM Studio, Google Gemini CLI, Mistral Vibe, OpenCode, Augment Code (Auggie CLI and Intent)
pip install hatch-xclamVerify:
hatch --versionOr install from source:
git clone https://github.com/CrackingShells/Hatch.git
cd Hatch
pip install -e .# Local server via npx — register it on VS Code
hatch mcp configure context7 --host vscode \
--command npx --args "-y @upstash/context7-mcp"
# Remote server with an auth header — register it on Gemini CLI
export GIT_PAT_TOKEN=your_github_personal_access_token
hatch mcp configure github-mcp --host gemini \
--httpUrl https://api.github.com/mcp \
--header Authorization="Bearer $GIT_PAT_TOKEN"
# Register the same server on multiple hosts at once
hatch mcp configure my-server --host claude-desktop,cursor,vscode \
--command python --args "-m my_server"# See all servers across all hosts
hatch mcp list servers
# See all hosts a specific server is registered on
hatch mcp show servers --server "context7"
# Detect which MCP host platforms are installed
hatch mcp discover hostsThe package system lets you install MCP servers with automatic dependency resolution and environment isolation. It is functional but being reworked for better integration with MCP registries.
hatch env create my_project
hatch env use my_project
hatch package add ./my_mcp_serverWe welcome contributions. See How to Contribute for details.
Quick setup:
- Fork and clone the repository
- Install dependencies:
pip install -e .andnpm install - Create a feature branch:
git checkout -b feat/your-feature - Make changes and add tests
- Use conventional commits:
npm run commit - Run tests:
wobble - Open a pull request
GNU Affero General Public License v3 — see LICENSE for details.
