capture

You can automatically capture knowledge from git commits by analyzing session changes across repos to surface nuggets, patterns, runbooks, and decisions as candidates for human review. The tool scans commits and changed files to generate these structured proposals.

Review-driven persistence works by writing pending capture items to ~/logs/captures/pending, then requiring human approval before saving them to the knowledge base. You can run a review command to approve or edit proposals before they are persisted.

Yes, you can generate runbooks and decisions from your current session by analyzing commits and changed files. The tool identifies recurring patterns and proposes structured captures, requiring human approval before persisting them to the knowledge base.

No external dependencies are required to capture session work into a knowledge base. The tool operates independently, using scripts to analyze commits and manage frontmatter-backed captures, writing pending items directly to your local logs directory.

You manage the knowledge life cycle using warm, cold, and icebox tiers. The tool supports archiving and promotion workflows, allowing you to organize and maintain captured session work as it progresses through different storage tiers.

The best way to reduce manual note-taking is to automatically generate knowledge capture candidates from daily session work. By scanning commits and recurring patterns, the tool surfaces structured proposals for human review, enabling faster knowledge dissemination.