Claude MarketplacesSubmit

Doctree Mcp

joesaby/doctree-mcp

Gives Claude a navigable tree over your markdown, CSV, and JSONL docs instead of dumping everything into context. Exposes BM25 search, table-of-contents traversal, and node-by-node content retrieval so the agent can drill down breadcrumb-style rather than guess from flat chunks. No vector DB, no embeddings, no LLM calls at index time. The bundled skills teach procedural knowledge: search first, inspect the outline, navigate to specific sections, retrieve what you need. Also includes wiki write tools with duplicate detection and schema validation if you want the agent maintaining runbooks. Works over stdio for local dev or deploys as Streamable HTTP for team use. Think research librarian behavior, not keyword lottery.

GitHub stars: 1
tools: 8
auth: Required
transport: STDIO
env vars: 4
last updated: Feb 21, 2026

Tools

8 tools

Public tool metadata for what this MCP can expose to an agent.

search_documentsBM25 keyword search + facet filters + glossary expansion (markdown · CSV · JSONL)

BM25 keyword search + facet filters + glossary expansion (markdown · CSV · JSONL)

No parameter schema in public metadata yet.

get_treeTable of contents — headings, word counts, summaries

Table of contents — headings, word counts, summaries

No parameter schema in public metadata yet.

get_node_contentFull text of a specific section by node ID

Full text of a specific section by node ID

No parameter schema in public metadata yet.

navigate_treeA section plus all descendants in one call

A section plus all descendants in one call

No parameter schema in public metadata yet.

lookup_rowO(1) exact-key lookup for structured data rows (e.g. PROJ-44)

O(1) exact-key lookup for structured data rows (e.g. PROJ-44)

No parameter schema in public metadata yet.

find_similarDuplicate detection with overlap ratios

Duplicate detection with overlap ratios

No parameter schema in public metadata yet.

draft_wiki_entryScaffold: suggested path, inferred frontmatter, glossary hits

Scaffold: suggested path, inferred frontmatter, glossary hits

No parameter schema in public metadata yet.

write_wiki_entryValidated write: path containment, schema, duplicate guards, dry-run

Validated write: path containment, schema, duplicate guards, dry-run

No parameter schema in public metadata yet.

Sponsors

Promote your product hereReach 125k+ visitors monthly Promote your product hereReach 125k+ visitors monthly

Agent CaffeineSponsoredClose your lid, come back to finished work

SnugSponsoredThe clipboard manager for the AI era Promote your product hereReach 125k+ visitors monthly

WaxSponsoredBlock anything with supervision, without erasing your iPhone Promote your product hereReach 125k+ visitors monthly

knobsSponsoredSee your site on any phone, any day, in any language

Configuration

DOCS_ROOT*

Path to your markdown repository root

DOCS_GLOB

Glob pattern for finding markdown files (default: **/*.md)

DOCS_ROOTS

Multiple weighted collections: ./docs:1.0,./api:0.8 (alternative to DOCS_ROOT)

GLOSSARY_PATH

Path to glossary.json for query expansion (default: $DOCS_ROOT/glossary.json)

Readme

doctree-mcp

Agentic document retrieval over markdown, CSV, and JSONL. BM25 + tree navigation via MCP — no vector DB, no embeddings, no LLM calls at index time.

The pitch: MCP provides the structural primitives (a navigable tree, BM25, glossary, row lookup). The bundled skills provide the procedural knowledge (how to walk that tree). Together the agent behaves like a trained research librarian — not a one-shot searcher. See The Skill + MCP Pattern.

Quick Start

Have docs already? Point a client at them:

# In your AI tool's MCP config — see docs/CLIENTS.md for per-tool snippets
{ "mcpServers": { "doctree": {
    "command": "bunx", "args": ["doctree-mcp"],
    "env": { "DOCS_ROOT": "./docs", "WIKI_WRITE": "1" }
} } }

Restart the tool → ask "search the docs for X" or invoke the doc-read prompt.

Starting fresh? Scaffold a Karpathy-style LLM wiki:

bunx doctree-mcp init          # configure current tool
bunx doctree-mcp init --all    # configure every supported client
bunx doctree-mcp init --dry-run

View the full README on GitHub

Details

Repository: github.com/joesaby/doctree-mcp
Categories: AI & LLM Tools, Documents & Knowledge, Search & Web Crawling
Package: doctree-mcp
Transport: STDIO
Auth: Required
Registry: MCP registry, active

Related AI & LLM Tools MCP servers

1

memexkioie/memexLocal MCP memory for coding agents. SQLite on your machine — preferences persist across chats.1stars

2

fish-bridge — Session Knowledge Graphmakeamouse/fish-bridge-mcpCompresses AI chat sessions into a typed knowledge graph. Local/offline via Ollama.1stars

3

Longmemmarerem/longmemPersistent cross-project memory for Cursor and Claude Code using local semantic search.1stars

4

Mcp Servermetrxbots/mcp-serverTrack AI agent costs, detect waste, optimize models, and prove ROI. 23 MCP tools across 10 domains.0stars

5

PinRAGndjordjevic/pinragMCP RAG: index PDFs, repos, YouTube, Discord, text; optional YouTube vision; query with citations.1stars

6

Aiglarenugehs/aiglareAudit AI/LLM features for governance guardrails: confidence, fallback, validation, human-in-loop.1stars