neuledge/context

What it solves

AI agents often rely on outdated training data, leading them to suggest deprecated APIs or incorrect syntax when libraries release new versions. Context provides AI agents with instant, up-to-date documentation by allowing them to fetch and query the latest docs locally.

How it works

Context operates as a Model Context Protocol (MCP) server. It uses a community-driven registry containing over 100 popular libraries. When an agent needs documentation, the server automatically searches the registry, downloads the documentation as a compact SQLite database, and performs full-text search (FTS5) locally on your machine. Users can also manually add documentation from git repositories, local directories, or websites using llms.txt files.

Who it’s for

Developers using MCP-compatible AI agents (such as Claude Code, Cursor, Zed, or GitHub Copilot) who need their AI to have accurate, version-specific knowledge of the frameworks and libraries they are using.

Highlights

• Plug and Play: Automatically downloads required docs from a community registry upon request.
• Local-First: Documentation is stored in local SQLite databases for sub-10ms query speeds, offline access, and privacy.
• Languge Agnostic: Supports documentation in Markdown, AsciiDoc, reStructuredText, and HTML across various ecosystems (npm, pip, maven).
• Extensible: Allows users to build custom documentation packages from private repos or internal websites.
• Broad Integration: Compatible with a wide range of AI IDEs and agents via the MCP standard.