skills.rest
nabitllc
nabitllcOFFICIAL
Explore all skills by this author →
GitHub

todero-retrospective

Generates a structured post-task retrospective proposing up to three skill line edits.

Updated Mar 26, 2026
Search Tags:#ai agents#retrospective#governance#lessons learned#skill improvement#postmortem#wrap-up
One-click install
npx skills add https://github.com/nabitllc/todero --skill todero-retrospective-nabitllc
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: todero-retrospective
Source: https://github.com/nabitllc/todero/tree/main/skills/todero-retrospective
Command: npx skills add https://github.com/nabitllc/todero --skill todero-retrospective-nabitllc

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? After finishing a task, lessons about what broke and what worked are usually lost. This Skill captures them immediately after wrap-up in a fixed retrospective format, turning mistakes into concrete, reviewable proposals for improving the skill pack. ## Core Features & Use Cases - Structured Three-Section Review: Records what went wrong (problem, detection, fix), what held (techniques worth keeping), and what to change (one to three proposed edits). - Proposal-Only Governance: Never edits skills directly; each proposal names the skill, quotes the exact line to change, and gives one sentence of why, for a human to apply and version-bump. - Use Case: After an AI agent completes a wrap-up on a Todero task, run this Skill to produce a retrospective noting that a triage skill's instruction caused a misclassification, quoting the offending line and proposing its replacement. ## Quick Start Run a retrospective on the task we just wrapped up and list up to three skill lines that should change.

Frequently Asked Questions about todero-retrospective

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

FAQPage Schema
How do I run a retrospective after an AI agent task?▼

Run the retrospective once, right after wrap-up while the work is still fresh. Record what went wrong and its fix, what held, and one to three proposed skill changes, each quoting the exact line and a one-sentence reason.

What should a post-task retrospective include?▼

Three sections: what went wrong with the problem, how it was noticed, and the fix; what held and which skill got it right; and what to change, naming the skill, quoting the line, and giving one sentence of why.

Can the retrospective edit skill files directly?▼

No. The skill only writes proposals. A person reviews each proposal, applies it manually, and bumps the target skill's version and last-synced date. That human review step is the only way a skill changes.

How many changes should a retrospective propose?▼

Between one and three entries. Three is a ceiling, not a target; if only one concrete lesson emerged, a single entry is correct. A retrospective is not finished until at least one concrete change is proposed.

When should the retrospective be skipped?▼

It is designed to run once per task immediately after wrap-up, not as a recurring weekly review. If nothing went wrong and nothing is worth changing, keep it short rather than inventing entries.

Related Skills

Explore More Software Engineering Skills→
gsd-buildgsd-build
7.8k

verify-before-complete

Enforces fresh verification output before any task completion claim.

Official
Intermediate
latecheckoutlatecheckout

update-skill

Update marketplace skills by bumping versions and validating JSON metadata.

Official
Intermediate
Yeachan-HeoYeachan-Heo
38.5k

ultraqa

Automate iterative QA cycles with tests, builds, lint, and type checks.

Community
Advanced
cursorcursor
6.6k

reflect

Spawns parallel review subagents over session transcripts to route learnings into skill edits.

Official
Advanced