Andrew Ng has announced the open-sourcing of Context Hub (chub), a command-line tool designed to help AI coding agents access structured documentation and retain learnings across tasks, marking a significant step in improving AI-assisted software development workflows.

The tool enables agents to fetch curated and versioned API documentation instead of relying on traditional web searches, thereby reducing common issues such as hallucinated APIs. It also allows agents to generate code and add annotations, including fixes, improvements, and edge cases, ensuring that valuable knowledge is preserved and reused across sessions.

Additionally, Context Hub incorporates a feedback system where agents can upvote or flag documentation, contributing to continuous improvement in content quality and reliability. This collaborative mechanism enhances the overall effectiveness of the platform.

Since its release, the project has rapidly scaled from fewer than 100 API documents to over 1,000 and has surpassed 6,000 GitHub stars within a week, reflecting strong developer interest and adoption.