Skip to content
Back to skills

Retro

ASecurity

Turn a shipped GTM branch into ledger lessons and guideline proposals; close resolved minors. Use when a branch has shipped.

  • 2 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added October 26, 2026
ai-agents

Security analysis

A100/100

Scanned November 11, 2026

npx -y skills add MorganKryze/GentleTeaM --skill retro --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Retro?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Retro
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/morgankryze-retro/badge)](https://www.skillsdirectory.com/skills/morgankryze-retro)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: retro
description: "Turn a shipped GTM branch into ledger lessons and guideline proposals; close resolved minors. Use when a branch has shipped."
argument-hint: "[slug]"
---
# Retro: make the next plan smarter

Ground rules: as in the `gtm` skill §0.

- Reflexes (${CLAUDE_PLUGIN_ROOT}/templates/methods/reflexes.md): Five Whys [brain], Hindsight
  Reflection [elicitation], Lessons Learned Extraction [elicitation].

## 1. Read
`<root>/reports/<slug>/` (progress, task reports, final review), the spec and plan, and the whole
ledger. Guideline edits follow `<root>/guidelines/voice.md`.

## 2. Distil
- **Lessons:** what would have saved a correction round, a blocked task or a final-review finding. One line
  each, phrased as a rule to keep (`- <date> · <slug> · lesson · <rule>`). At most five; fewer is
  better. Skip what a test or a lint rule now enforces. A lesson a reflex found ends with its
  lens tag, `(lens: <method name>)`: tune counts it.
- **Guidelines:** a lesson seen in two branches or more belongs in a guideline. Propose the edit
  (file, the line to add) as a decision card; apply on the human's answer.
- **Minors:** strike through (`~~ ~~`) the ones fixed since, with a `resolved` line naming them.
- **Superseded:** strike entries a newer one contradicts; never delete.
- **Sheets:** count, per sheet, across the `sheets` segments of every `run` line not struck through:
  runs, passes, findings by severity. A line without the segment counts for no sheet. A sheet with
  0 findings of any severity over the last 3 runs that name it gets one decision card: remove it or
  keep it; with fewer than 3 such runs, retro proposes nothing. A sheet lives in the plugin:
  retro never edits or deletes it or its index row; its removal is the human's, a reviewed commit
  there. Either answer is a `ruling` line naming the sheet and quoting the answer; a kept sheet is
  proposed again only after 3 runs that name it, newer than that line.
  A sheet with a standing remove ruling, or with no row in
  `${CLAUDE_PLUGIN_ROOT}/templates/sheets/index.md`, gets no card.

## 3. Spot patterns for /gtm:tune
Suggest `/gtm:tune` in one line when the `run` lines show: correction rounds hitting the cap in two runs
or more; a specialist that found nothing across three runs it ran on; a final-review finding a
specialist should have caught; design loops ending at the last round; `usage` notes about limits.

## 4. Report
TL;DR (lessons added, guideline proposals, minors closed), the sheets' counts in one table (sheet,
runs, passes, findings `c/i/m`) when a line has them, then the decision cards, if any.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

View all in ai-agents →

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.