Ai-Sessions

Tracks which lines of code were written by AI vs humans, storing attribution logs as git notes that survive rebases, squashes, and cherry-picks

Attaches cleaned markdown conversation transcripts to commits using native git notes – the lightest-touch approach to AI session archival

Exports Claude Code JSONL session transcripts to human-readable markdown and structured JSON for archival and analysis

Captures Claude Code sessions from local JSONL storage and produces structured transcripts for archival in git repositories

Use Claude Code’s PreCompact, Stop, and SessionEnd hooks to trigger automatic archival of AI coding session transcripts into git repositories

Stores AI coding session transcripts on separate git shadow branches, preserving full provenance without polluting the working tree

Saves AI coding sessions from VS Code and Cursor as structured markdown files in a .specstory/ directory within the project