retro

Generates sprint retrospectives from ledger facts including estimates, model mix, and waste.

1|Updated Aug 17, 2026
One-click install
npx skills add https://github.com/Jakeintech/waybill --skill retro-jakeintech
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: retro
Source: https://github.com/Jakeintech/waybill/tree/main/skills/retro
Command: npx skills add https://github.com/Jakeintech/waybill --skill retro-jakeintech

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Retrospectives run from memory rehash opinions instead of facts. This Skill renders a sprint retro pack directly from the waybill ledger, so shipped work, estimate accuracy, token costs, and waste are grounded in metered data rather than recollection. ## Core Features & Use Cases - Shipped work summary: Lists each shipped item with points, PRs, tokens, and cost from the ledger report. - Estimate calibration: Compares pre-registered estimates against actuals, plainly naming items that exceeded the no-Claude estimate range. - Model mix and waste analysis: Reports tokens-per-point by dominant model, retried commands, reopened items, and cache savings derived from the spend ledger. - Use Case: At the end of a two-week sprint, ask for a retro and receive a fact-based pack showing which estimates held, which model carried which stories, and which items sat idle with spend attached. ## Quick Start Ask the assistant to run a retro for your last sprint so it queries the waybill ledger and renders the fact-based retro pack.

Frequently Asked Questions about retro

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run a sprint retrospective with real data?▼

Ask for a retro and name the sprint window. The skill runs waybill mine to catch up metering, then queries the report, manifest, and untracked endpoints to render shipped work, calibration, model mix, waste, and idle items from ledger facts.

How does estimate calibration work in a retro?▼

Calibration compares pre-registered estimates against actuals for shipped items. Items below the no-Claude range held the claimed saving, items within were partial, and items above took longer than estimated and are named plainly rather than softened.

What time window does the retro use by default?▼

The window is the sprint the user names. If they say last sprint, it uses the sprint length from config.baseline.window when known, otherwise the last 14 days, and the chosen window is always stated in the header.

Does the retro skill write anything to the ledger?▼

No. The skill is read-only: it runs metering catch-up and queries, then renders the pack. It never writes ledger entries, and any user-supplied context is included marked as theirs.

Can I share the retro output with my team or leadership?▼

Yes, but re-run the queries with the --audience internal or --audience external flag rather than hand-editing the output. The skill never includes peer comparisons, only facts about the user's own work.