context7

Exposes documentation for 30+ library versions through the Model Context Protocol (MCP) standard, implementing a two-tool system (resolve-library-id and query-docs) that maps natural language library references to specific versions and retrieves ranked, semantically-relevant documentation snippets. The system uses LLM-powered ranking to surface the most contextually relevant documentation sections rather than simple keyword matching, enabling AI assistants to access current API signatures and examples without hallucination.

The resolve-library-id MCP tool automatically maps natural language library references (e.g., 'React', 'the HTTP client I'm using') to specific library identifiers and versions by analyzing the developer's codebase context and project dependencies. This capability eliminates the need for explicit version specification by examining package.json, import statements, and AI editor context to infer which version the developer is actually using.

Context7 provides APIs and workflows for adding custom libraries to its documentation index, including automatic documentation parsing, version tracking, and indexing for semantic search. The system supports adding libraries via REST API endpoints, CLI commands, or web dashboard, with support for multiple documentation formats (Markdown, HTML, JSDoc) and automatic version detection from package manifests.

Context7 provides a web dashboard for managing libraries, viewing usage metrics, configuring teamspaces, and managing billing. The dashboard displays documentation lookup statistics, API usage, team member access, and library management controls, enabling teams to monitor documentation usage patterns and manage access across multiple developers.

Context7 supports enterprise on-premise deployment via Docker Compose and Kubernetes, enabling organizations to run the entire documentation service within their own infrastructure. The deployment includes support for private documentation storage, custom authentication (OAuth 2.0, SAML), and teamspace policies for managing access across departments.

Context7 provides a Docs Researcher Agent that autonomously discovers and fetches relevant documentation based on developer queries or code context, automatically injecting documentation into the AI assistant's context without explicit user invocation. The agent uses auto-invoke rules to detect when documentation might be relevant and proactively fetches it, reducing the need for manual documentation lookup.

Context7 implements the Model Context Protocol (MCP) specification to expose documentation tools through a standardized interface that works across 30+ AI coding assistants (Cursor, Claude Code, VS Code Copilot, Windsurf, etc.) without requiring separate integrations for each client. The MCP server exposes tools via stdio, HTTP, or SSE transports, allowing clients to discover and invoke documentation retrieval with consistent schemas and error handling.

The query-docs MCP tool implements semantic search over indexed library documentation using LLM-powered ranking that understands developer intent and filters results by library version. Rather than keyword matching, the system uses embeddings and LLM-based relevance scoring to surface documentation sections that are semantically related to the developer's query, with results ranked by relevance to the specific library version being used.

The ctx7 CLI tool provides developers with command-line access to documentation lookup, library management, and skill generation workflows. It enables local documentation queries without requiring an AI editor, supports adding custom libraries to the Context7 index, manages authentication tokens, and generates reusable 'skills' (documentation snapshots) that can be shared across teams or embedded in prompts.

Context7 provides native plugins for Cursor and Claude Code that integrate documentation retrieval directly into the editor's UI and chat interface. The plugins support auto-invoke rules (automatically trigger documentation lookup based on patterns like '@library-name' mentions), custom skills (reusable documentation snapshots), and agent-based documentation research that can autonomously fetch relevant docs without explicit user invocation.

Context7 provides a TypeScript SDK (@upstash/context7-sdk) that enables developers to programmatically query documentation, manage libraries, and integrate documentation retrieval into custom applications, agents, or workflows. The SDK wraps the REST API with type-safe methods, supports both Node.js and browser environments, and includes utilities for embedding documentation in LLM prompts.

Context7 provides integration with Vercel's AI SDK through the @upstash/context7-tools-ai-sdk package, enabling developers to define documentation retrieval as a tool within AI SDK-based applications. The integration automatically handles tool schema generation, parameter validation, and result formatting, allowing documentation lookup to be seamlessly integrated into AI SDK's tool-calling workflows.

Context7 supports remote MCP server deployment via HTTP and Server-Sent Events (SSE) transports, enabling clients to connect to a centralized documentation server without running a local instance. The deployment includes built-in authentication (API key validation), rate limiting, and support for Docker and Kubernetes deployments, allowing teams to manage a single shared documentation service.

Context7 supports local MCP server deployment via stdio transport, enabling developers to run a documentation server on their local machine with direct process communication to their AI editor. This approach eliminates network latency and provides offline documentation access, with automatic server lifecycle management integrated into editor configuration.