mcp
MCP server from cressie176/mcp-server
claude mcp add --transport stdio cressie176-mcp-server npx --yes github:cressie176/mcp-server --repository-type github --user your-username --repository your-repo
How to use
This MCP server provides access to resources and prompts stored in a GitHub repository (or a local filesystem for testing). It reads an index.json that declares available resources and prompts, then serves the corresponding markdown content to Claude and other MCP-compatible tools. With the default setup, you point Claude at a repository, and the server exposes the defined resources and prompts via the MCP interface so that you can browse, list, and prompt the model with structured guidance. The examples show how to run the server via npx and how to reference the repository, user, and repository path so Claude can fetch the correct assets. You can also run the server against a local filesystem to test changes before pushing to GitHub, using the filesystem path option described in the quick start and usage sections.
How to install
Prerequisites:
- Node.js and npm installed on your machine (npx is included with npm).
- A GitHub repository (or local directory) following the required structure.
-
Verify Node.js and npm are installed:
node -v npm -v
-
Prepare your repository:
-
Create a repo with the following structure: your-repo/ ├── index.json ├── resources/ │ └── your-resource.md └── prompts/ └── your-prompt.md
-
Create index.json that defines resources and prompts, for example: { "resources": [{"name": "your-resource", "description": "Description of your resource"}], "prompts": [{"name": "your-prompt", "description": "Description of your prompt"}] }
-
-
Run the MCP server via npx (GitHub-hosted):
npx --yes github:cressie176/mcp-server --repository-type github --user your-username --repository your-repo
If you want to test with a local filesystem instead of GitHub:
npx --yes github:cressie176/mcp-server --repository-type filesystem --path /path/to/your/data
-
Optional: add the command to your Claude configuration or your automation script to restore the server easily.
-
Validate that Claude can access the resources with the /mcp command or the equivalent API endpoints provided by MCP.
Additional notes
Tips and common issues:
- Ensure index.json is valid JSON and correctly references the resource and prompt files in resources/ and prompts/.
- When using GitHub, provide correct values for --user and --repository; ensure the repository contains index.json and the expected directory structure.
- If you change repository paths or filenames, update index.json accordingly.
- For local filesystem testing, make sure the path points to a directory containing index.json, resources/, and prompts/ with the same structure.
- Log level and log file options are exposed by the underlying MCP server; use --log-level to adjust verbosity and --log-file to persist logs if needed.
- If you encounter MIME or encoding issues with Markdown, ensure resource/prompts are valid Markdown (.md) files.
- The server can be restarted after changes to the repository to refresh the available resources and prompts in Claude.
Related MCP Servers
zen
Selfhosted notes app. Single golang binary, notes stored as markdown within SQLite, full-text search, very low resource usage
MCP -Deepseek_R1
A Model Context Protocol (MCP) server implementation connecting Claude Desktop with DeepSeek's language models (R1/V3)
mcp-fhir
A Model Context Protocol implementation for FHIR
mcp
Inkdrop Model Context Protocol Server
mcp-appium-gestures
This is a Model Context Protocol (MCP) server providing resources and tools for Appium mobile gestures using Actions API..
dubco -npm
The (Unofficial) dubco-mcp-server enables AI assistants to manage Dub.co short links via the Model Context Protocol. It provides three MCP tools: create_link for generating new short URLs, update_link for modifying existing links, and delete_link for removing short links.