What
Adds growth-retrospective to the Productivity & Organization category.
Why this category
It's a personal-growth retrospective tool — closer to ship-learn-next and tapestry (already in this section) than to the Collaboration / Project Management section. Solo dev / indie builder is the target user.
What's different from existing entries in this section
- Graduation markers: every Tier 1/2 item ships with an observable, time-bounded exit condition. Items leave the list — the list shrinks as you actually improve.
ship-learn-next and tapestry don't have this exit-criteria semantic.
- Step 0 Action Audit (forcing function): before any new scan, the skill checks the filesystem for artifacts named in prior level-up actions. If ≥3 are pending >14 days AND 0 completed, the skill pushes back instead of generating new items. Designed to prevent the "items pile up forever" failure mode common to learning logs.
- 5-category sweep (domain / human / habit / meta / productivity) — most learning tools track only category 1.
- Cross-agent: same SKILL.md format works in Claude Code, OpenAI Codex CLI, and Google Antigravity.
Requirements checklist
Try it in 5 seconds without installing
git clone https://github.com/YoungApple/growth-retrospective-skill /tmp/grs && \
bash /tmp/grs/examples/demo-fixture/demo.sh
Color-coded output of Step 0 Action Audit + signal extraction on a synthetic 14-day project. Pure shell, no LLM, no install. About 5 seconds.
Honest notes
The "shrinking gap" outcome claim is Layer 3 in the eval framework and currently has zero real-user data — see EVAL-FRAMEWORK.md. Layers 1 (output) and 2 (acceptance via persona simulation) are validated.
Brand new repo (created 2026-05-24, 1 star at time of submission). If the policy is "wait for community traction before listing," I understand a hold — but the CONTRIBUTING.md doesn't specify a stars threshold and emphasizes merit / use-case / structure, which I believe this skill meets.