feature-miner

Session mining scans indexed coding agent conversations to identify documentation usage patterns, staleness, and friction. It expands high-scoring hits for context and synthesizes signals into categorized findings for autodoc prioritization.

To identify stale content and frontmatter issues, run parallel searches across cass-indexed agent sessions. The analysis expands high-scoring hits for surrounding context and synthesizes signals into categorized findings within a markdown report.

Yes, you need a healthy cass CLI to query workspaces and python3 to merge and analyze results. These dependencies are required to produce the consolidated JSON results file and the human-readable markdown report.

The best way to extract feature ideas is to scan historical agent sessions for discovery failures and CLI friction. Running targeted searches and expanding contextual hits synthesizes signals into ranked findings and recommended next steps.

Yes, you can apply session mining to cass-indexed sessions across workspaces and agent types. It produces a human-readable markdown report with ranked findings, pain points, and recommended next steps specifically for autodoc prioritization.