
jDocMunch MCP
Supports UIby jgravelle
Token-efficient documentation exploration via structured section indexing, preventing brute-force file reading.
What it does
Indexes documentation sets by heading hierarchy and section structure, allowing AI agents to retrieve precise sections of a document instead of reading entire files, significantly reducing token consumption.
Tools
index_local: Indexes a local documentation folder.index_repo: Indexes a GitHub repository's documentation.search_sections: Performs weighted search returning summaries of documentation sections.get_section: Retrieves the full content of a specific section using its stable ID.get_toc: Returns a flat section list in document order.get_toc_tree: Returns a nested section tree per document.get_document_outline: Returns the section hierarchy for a single document.get_sections: Performs batch content retrieval for multiple sections.delete_index: Removes a documentation index.get_broken_links: Detects internal links or anchors that no longer resolve.get_doc_coverage: Maps documentation sections to code symbols.
Installation
Using uvx (recommended):
{
"mcpServers": {
"jdocmunch": {
"command": "uvx",
"args": ["jdocmunch-mcp"]
}
}
}
Supported hosts
Confirmed support for Claude Desktop, Claude Code, and OpenClaw.
Quick install
pip install jdocmunch-mcpInformation
- Pricing
- freemium
- Published






