andrewyng / context-hub

Context Hub Coding agents hallucinate APIs and forget what they learn in a session. Context Hub gives them curated, versioned docs, plus the ability to get smarter with every task. All content is open and maintained as markdown in this repo — you can inspect exactly what your agent reads, and contribute back. Quick Start How It Works Chub is designed for your coding agent to use (not for you to use!). You can prompt your agent to use it (e.g., "Use the CLI command chub to get the latest API documentation for calling OpenAI. Run 'chub help' to understand how it works.") Or by creating an agent skill to use Chub using SKILL.md, and ideally prompting your agent to remember to use this skill. (If you are using Claude Code, create the directory ~/.claude/skills/get-api-docs and put SKILL.md there.) **Most of the time, it's simple — search, fetch, use:** **When the agent discovers a gap**, it can annotate locally for next time: **Feedback flows back to authors** — or — vote the docs up or down so they can get better for everyone over time. Content Types Versioned, language-specific. "What to know." More content types than API documentation (such as agent skills) are on the roadmap. Commands | Command | Purpose | |---------|---------| | | Search docs and skills (no query = list all) | | | Fetch docs or skills by ID | | | Attach a note to a doc or skill | | | Remove annotations | | | List all annotations | | | Upvote or downvote a doc (sent to maintainers) | For the full list of commands, flags, and piping patterns, see the CLI Reference. Self-Improving Agents Context Hub is designed for a loop where agents get better over time. **Annotations** are local notes that agents attach to docs. They persist across sessions and appear automatically on future fetches — so agents learn from past experience. See Feedback and Annotations. **Feedback** (up/down ratings with optional labels) goes to doc authors, who update the content based on what's working and what isn't. The docs get better for everyone — not just your local annotations. Key Features Incremental Fetch Docs can have multiple reference files beyond the main entry point. Fetch only what you need — no wasted tokens. Use to grab specific references, or for everything. See the CLI Reference. Annotations & Feedback Annotations are local notes that agents attach to docs — they persist across sessions and appear automatically on future fetches. Feedback (up/down ratings) goes to doc authors to improve the content for everyone. See Feedback and Annotations. Contributing Anyone can contribute docs and skills — API providers, framework authors, and the community. Content is plain markdown with YAML frontmatter, submitted as pull requests. See the Content Guide for the format and structure. Agent feedback (up/down ratings from real usage) flows back to authors, helping surface what needs fixing and improving overall quality over time. License MIT