report

What problem does it solve? Turning AI-assisted coding work into a credible summary for managers, reviewers, or token-budget decision-makers usually means hand-assembling numbers that nobody can verify. This Skill builds formal one-page reports — token pitches, performance reviews, sprint recaps, quarterly summaries, grant reports — directly from the waybill ledger, so every figure is metered from local transcripts and git history rather than estimated. ## Core Features & Use Cases - Preset report types: token-pitch, perf-review, sprint-recap, quarterly, grant-report, and incident windows, each with its own structure, audience redaction level, and section budget. - Engine-sourced numbers: all metrics come from waybill query report / query spend output — ranges stay ranges, unattributed spend is always shown, and nothing is recomputed or padded. - Bootstrap mode: when the ledger has fewer than three shipped entries, produces a facts-only report from waybill bootstrap instead of fabricating claims. - Verification pack: optionally exports the raw event lines and engine so a recipient can re-run conservation checks offline. - Use Case: Ask for a "quarterly Claude report" and receive a one-page artifact listing shipped items with PR links, tokens-per-point efficiency, spend with confidence labels, and a methodology footnote — ready to paste into a review packet. ## Quick Start Ask the assistant to generate your token pitch or quarterly accomplishment report from the waybill ledger for the period you care about.