cc-insights

Semantic search uses sentence-transformers to convert conversations into vector embeddings stored in ChromaDB, enabling you to find relevant discussions by intent rather than exact phrase matches. This skill indexes your project histories automatically and supports fast retrieval by topic.

Yes, this skill indexes conversations from ~/.claude/projects incrementally and detects patterns to generate automated insights, including weekly reports, file heatmaps, and trend analysis that surface which files and topics dominate your interactions.

File heatmaps visualize which files appear most frequently across your conversations, helping you spot areas needing attention. The skill generates these automatically during analysis using incremental indexing of your project histories.

Indexing is incremental and automated. The skill processes conversations from your project directory, updates the vector database, and prepares data for search and reporting without requiring manual intervention between runs.

Yes, this skill generates weekly insight reports using Jinja2 templating and pattern detection. Reports extract trends and actionable findings from indexed conversations, formatted for dashboard integration and team sharing.

You need Claude Code conversations stored in ~/.claude/projects. The skill handles environment setup including ChromaDB initialization, sentence-transformer model loading, and incremental indexing—no manual database configuration required.