MCP Documentation Management Service - A Model Context Protocol implementation for documentation management
Configuration
View docs{
"mcpServers": {
"alekspetrov-mcp-docs-service": {
"command": "npx",
"args": [
"-y",
"mcp-docs-service",
"/path/to/your/docs"
],
"env": {
"MCP_NPX_WRAPPER": "true"
}
}
}
}You can manage your Markdown documentation with an MCP server that reads, writes, edits, and analyzes frontmatter-enabled documents, making it easy to work with AI assistants and maintain high-quality docs across your projects.
You will connect a client to the Docs Manager MCP server to perform common documentation tasks. Use natural language prompts to read, search, or check health, or use explicit tool commands when you need precise control. The server is designed to work with standard Markdown files that include YAML frontmatter and to produce outputs optimized for large language models.
Prerequisites: you need Node.js installed on your machine.
Install the MCP Docs Service globally using npm:
npm install -g mcp-docs-serviceOr run it directly with npx, passing the path to your documentation directory:
npx mcp-docs-service /path/to/docsCursor integration relies on a project-specific configuration. Create a Cursor integration snippet in your project root to expose the MCP Docs Service as a tool named docs-manager:
{
"mcpServers": {
"docs-manager": {
"command": "npx",
"args": ["-y", "mcp-docs-service", "/path/to/your/docs"]
}
}
}Claude Desktop integration is provided via a similar configuration. Ensure you place the MCP server under the docs-manager key and, if required, set the MCP_NPX_WRAPPER environment variable when starting Claude Desktop. Restart Claude Desktop after updating the configuration.
Note: when using the CLI or the browser-based tooling, always point to the absolute path of your documentation directory to avoid path resolution issues.
Configuration and health are built into the service. You can perform health checks to identify missing metadata, broken links, or orphaned documents, and you can generate a consolidated single-document output tailored for large language models.
The service is resilient by default. It returns a minimum health score even if issues exist, creates missing directories automatically, and continues processing when encountering errors. This helps you gradually improve your documentation without blocking progress.