context7

Implements a Model Context Protocol server that exposes documentation as callable tools for 30+ AI coding assistants (Cursor, Claude Code, VS Code Copilot, Windsurf). Uses an indexed, searchable documentation store with LLM-powered ranking to surface the most relevant library documentation snippets for a given query, preventing API hallucinations by grounding LLM responses in current, version-specific docs. The MCP transport layer abstracts away client-specific integration details, allowing a single server implementation to serve multiple AI editor ecosystems.

Implements the 'resolve-library-id' MCP tool that automatically identifies which libraries are referenced in code or natural language queries, then resolves them to canonical library identifiers in Context7's index. Uses pattern matching, import statement parsing, and semantic understanding to handle aliases, monorepo packages, and version specifiers. The tool bridges the 'Natural Language Space' of developer prompts to the 'Code Entity Space' of indexed libraries, enabling downstream documentation queries without explicit library name specification.

Provides a web dashboard for monitoring Context7 usage, viewing query history, managing team access, and configuring library settings. Includes usage metrics (queries/month, libraries accessed, top queries), teamspace management (invite team members, set permissions), and library admin panel (claim libraries, manage documentation, view indexing status). Supports OAuth 2.0 for authentication and role-based access control (admin, editor, viewer). Analytics data is aggregated and anonymized for privacy.

Provides enterprise-grade deployment options including on-premise Docker Compose setup, Kubernetes deployment with Helm charts, and managed cloud deployment. Supports private repository access for internal libraries, custom authentication (OAuth 2.0, LDAP, SAML), and data residency compliance (GDPR, HIPAA). Includes Docker Compose templates for single-server deployment and Kubernetes manifests for multi-node clusters. Enterprise plans include SLA guarantees, dedicated support, and custom rate limits.

Provides a GitHub Action that integrates Context7 into CI/CD pipelines for automated documentation validation. The action can query documentation for dependencies, validate generated code against official docs, and fail builds if documentation is outdated or unavailable. Supports matrix builds for testing against multiple library versions. Outputs validation results as GitHub check annotations and workflow artifacts. Can be combined with CodeRabbit integration for code review automation.

Implements the 'query-docs' MCP tool that accepts natural language queries and returns ranked documentation snippets from the indexed library store. Uses semantic search (embeddings-based) combined with LLM-powered re-ranking to surface the most contextually relevant documentation. The ranking algorithm considers query intent, code context, library version, and documentation freshness. Results are returned with source attribution and version metadata, enabling LLMs to cite specific documentation sources.

Provides a Model Context Protocol server implementation that abstracts away client-specific integration details, allowing a single codebase to serve Cursor, Claude Code, VS Code Copilot, Windsurf, and other MCP-compatible clients. Supports both remote deployment (at mcp.context7.com) and local deployment (Docker, Kubernetes, on-premise). The transport layer handles stdio, HTTP, and WebSocket protocols transparently. Configuration is client-specific (via ctx7 CLI setup command or manual config files), but the core MCP tool definitions remain consistent across all clients.

Implements a documentation ingestion pipeline that crawls library documentation (from npm, GitHub, official docs sites), parses it into semantic chunks, generates embeddings, and stores them with version metadata. The system maintains a searchable index of 1000+ libraries with version-specific documentation. Supports manual library registration via the Context7 admin panel for private or custom packages. The indexing process includes deduplication, freshness tracking, and LLM-powered summarization of documentation sections for improved ranking.

Exposes Context7's documentation retrieval capabilities via REST endpoints at https://mcp.context7.com/mcp, enabling programmatic access from non-MCP clients (scripts, CI/CD pipelines, custom integrations). Implements OAuth 2.0 and API key authentication with per-plan rate limits (typically 100-1000 queries/month). Endpoints include search-libraries (find available libraries), get-context (retrieve documentation for a library), and add-library (register custom libraries). Responses are JSON-formatted with version metadata and source attribution.

Provides a command-line interface for querying documentation, managing libraries, and configuring Context7 integration. Commands include 'ctx7 docs' for querying documentation, 'ctx7 add-library' for registering custom libraries, 'ctx7 auth' for authentication setup, and 'ctx7 setup' for MCP client configuration. The CLI uses the TypeScript SDK internally and outputs results in human-readable format (Markdown, JSON). Supports shell completion and integration with shell scripts and CI/CD pipelines.

Provides @upstash/context7-sdk and @upstash/context7-tools-ai-sdk npm packages for direct TypeScript/JavaScript integration without MCP. The SDK exposes methods for querying documentation (query-docs), resolving libraries (resolve-library-id), and managing libraries. Supports both Node.js and browser environments. The AI SDK package includes integration with Vercel's AI SDK, enabling use in AI agent frameworks. SDK handles authentication, error handling, and response parsing internally.

Provides native plugins for Cursor, Claude Code, and VS Code that integrate Context7 documentation retrieval directly into editor workflows. Plugins expose documentation queries through editor commands, context menus, and auto-invoke rules (rules that automatically trigger documentation retrieval based on code patterns). Cursor and Claude Code plugins include 'Skills' (reusable documentation-aware prompts) and 'Agents' (multi-step documentation workflows). VS Code plugin uses rules and skills for configuration. Auto-invoke rules enable automatic documentation retrieval when developers mention specific libraries or write code patterns.

Implements an autonomous agent that automatically identifies libraries in code or prompts, retrieves relevant documentation, and synthesizes answers to developer questions. The agent uses a multi-step reasoning loop: identify libraries (via resolve-library-id), query documentation (via query-docs), rank results by relevance, and generate explanations grounded in official docs. Supports auto-invoke rules that trigger the agent when developers mention specific libraries or write code patterns. The agent maintains context across multiple turns, enabling follow-up questions about documentation.