What problem does it solve? After finishing an epic, teams rarely know what the delivery actually cost in tokens, cache usage, money, and code volume. This Skill collects those statistics from local session transcripts, git history, and the issue tracker, then produces an HTML dashboard and a session summary without sending data anywhere. ## Core Features & Use Cases - Epic-level cost reporting: Aggregates tokens by model, cache efficiency, estimated cost, subscription window, and code volume for an epic and all its child tickets. - Baseline comparison: Saves a versioned baseline from a completed epic and compares it against the next one, preserving exact versus estimated attribution per provider. - Honest attribution caveats: Distinguishes exactly-attributed Claude Code work from approximately-attributed Codex work, and reports missing data as "no data" instead of zero. - Use Case: After closing epic #42, run the tool to produce docs/reports/epic-42.html showing per-model token spend, cache split, and cost priced from your own rates.json, then compare it against the baseline saved from epic #38. ## Quick Start Ask the agent to run the delivery stats report for your finished epic, for example: generate the delivery statistics dashboard for epic 42 and save it under docs/reports.
