MCP Tools Reference

This page lists the MCP tools exposed by Reader's standalone MCP server.

Source of truth:

• mcp-server/src/tools/markdown-tools.mjs

Tool List

• reader.list_documents
• reader.open_document
• reader.get_markdown_outline
• reader.search_markdown
• reader.semantic_search_documents
• reader.import_document

Compatibility aliases accepted by tools/call:

• reader.list_markdown_documents -> reader.list_documents
• reader.open_markdown_document -> reader.open_document

1) reader.list_documents

Description: List documents from Reader SQLite (markdown / pdf / epub).

Input schema summary:

• db_path?: string optional DB path override
• limit?: integer range 1..500, default 100
• file_types?: string[] enum values: markdown, pdf, epub

Example:

{
  "limit": 50,
  "file_types": ["markdown", "pdf"]
}

2) reader.open_document

Description: Read document content by one locator.

Locator requirement (any one):

• doc_id
• path
• title

Input schema summary:

• db_path?: string
• doc_id?: string
• path?: string
• title?: string
• max_chars?: integer range 200..400000, default 12000

Example:

{
  "doc_id": "doc_xxx",
  "max_chars": 16000
}

3) reader.get_markdown_outline

Description: Return section outline for a target document.

Locator requirement (any one):

• doc_id
• path
• title

Input schema summary:

• db_path?: string
• doc_id?: string
• path?: string
• title?: string

Example:

{
  "title": "Deep Learning with Python"
}

4) reader.search_markdown

Description: Keyword search in paragraph content, optional document scope.

Input schema summary:

• db_path?: string
• doc_id?: string
• path?: string
• title?: string
• query: string required
• case_sensitive?: boolean default false
• limit?: integer range 1..200, default 20

Example:

{
  "query": "transformer",
  "limit": 20,
  "case_sensitive": false
}

5) reader.semantic_search_documents

Description: Cross-document vector similarity search over embedding index.

Required condition (any one):

• query (natural language)
• query_vector (precomputed embedding)

Input schema summary:

• db_path?: string
• query?: string
• query_vector?: number[]
• top_k?: integer range 1..200, default 10
• min_score?: number range -1..1, default -1
• scan_limit?: integer range 10..200000, default 20000
• batch_size?: integer range 200..5000, default 2000
• doc_id?: string
• path?: string
• title?: string
• file_types?: string[] enum: markdown, pdf, epub
• embedding_profile?: { provider?: string; model?: string; dimension?: integer }
• embedding_provider?: string
• embedding_model?: string
• embedding_dimension?: integer
• lm_studio_url?: string
• openai_base_url?: string
• openai_api_key?: string
• embedding_ollama_url?: string
• embedding_ollama_model?: string
• local_model_path?: string

Example:

{
  "query": "what are the key claims in chapter 3",
  "top_k": 8,
  "file_types": ["markdown"]
}

6) reader.import_document

Description: Import local markdown/pdf/epub file into Reader database.

Input schema summary:

• db_path?: string
• path: string required
• title?: string
• author?: string
• language?: string
• force_reimport?: boolean default false

Example:

{
  "path": "/Users/me/Documents/book.epub",
  "force_reimport": true
}

MCP Resources (also exposed)

• reader://project
• reader://database
• reader://documents/recent