@notionhq/notion-mcp-server

Exposes Notion database querying through the Model Context Protocol, allowing AI agents and LLM applications to execute structured queries against Notion databases without direct API calls. Implements MCP resource handlers that translate database queries into Notion API calls, returning paginated results with full property metadata and filtering support.

Retrieves full page content from Notion including nested block structures (paragraphs, headings, lists, code blocks, tables) and parses them into structured format. Implements recursive block traversal to handle Notion's hierarchical block model, converting rich text formatting, mentions, and embedded content into accessible text representations for LLM consumption.

Creates new pages in Notion databases with full property assignment through MCP tool calls. Implements property type mapping (text, select, multi-select, date, checkbox, relations) to convert LLM-generated values into Notion's property schema format, handling type validation and enum constraints before API submission.

Updates existing Notion pages and modifies their properties through MCP tool calls. Implements partial update semantics where only specified properties are changed, with conflict detection and type validation. Supports updating rich text content, property values, and page metadata while preserving unmodified fields.

Exposes Notion database schema through MCP resources, allowing AI agents to discover available properties, their types, constraints (enums, date formats), and relationships. Implements schema caching to reduce API calls and provides property metadata needed for intelligent form generation or validation in downstream systems.

Implements the Model Context Protocol server specification, handling bidirectional JSON-RPC communication with MCP clients, request routing, and authentication token management. Manages Notion API credentials securely, refreshing tokens as needed and abstracting authentication details from client implementations.

Tracks and exposes the authenticated user context and their permissions within Notion workspaces through MCP. Provides information about which pages and databases the authenticated user can access, enabling permission-aware operations and preventing unauthorized access attempts before they reach the Notion API.

Implements full-text search across Notion workspaces through MCP, allowing AI agents to find pages and database records by content or title. Leverages Notion's search API to return ranked results with relevance scoring, enabling semantic knowledge retrieval without requiring external vector databases or indexing infrastructure.