retro

Extract evidence-backed lessons from development sessions and propose repository guidance improvements.

1.3k|142|Updated Jun 1, 2026
One-click install
npx skills add https://github.com/MiniMax-AI/minimax-code --skill retro-minimax-ai
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: retro
Source: https://github.com/MiniMax-AI/minimax-code/tree/main/.agents/skills/retro
Command: npx skills add https://github.com/MiniMax-AI/minimax-code --skill retro-minimax-ai

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Development sessions often surface failures, retries, and workflow gaps that get forgotten instead of being converted into lasting improvements. This Skill turns those concrete events into a small set of evidence-backed lessons and proposes focused updates to repository guidance such as AGENTS.md, CONTRIBUTING.md, or docs. ## Core Features & Use Cases - Evidence-Based Lesson Extraction: Records each lesson with the event, demonstrated cause, recommended change, and impact category, while skipping unsupported opinions and one-off circumstances. - Guidance Placement: Maps each lesson to the right destination, whether repository-wide rules, a specific skill entry, docs procedures, or a code fix with a regression test. - Scoped Delivery: Returns findings and proposed edits by default, applies documentation changes only when requested, and redacts credentials and private details from evidence. - Use Case: After a long debugging session where a stale build script caused repeated test failures, run a retrospective to capture the root cause and propose a corrected build procedure in docs plus a regression test suggestion. ## Quick Start Ask the assistant to run a retrospective on the completed development session and propose improvements to the repository guidance.

Frequently Asked Questions about retro

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

FAQPage Schema
How do I run a retrospective on a development session?

Ask for a retrospective after a completed session and the Skill reviews failures, retries, and workflow gaps. It records each lesson with the event, cause, recommended change, and impact, then proposes where the improvement belongs.

How to capture reusable lessons from coding failures?

Each lesson is tied to concrete evidence such as a file, command, error, or commit, with the demonstrated cause separated from hypotheses. Duplicate lessons are merged and one-off circumstances are excluded so only reusable guidance remains.

Where should retrospective lessons be documented in a repository?

Repository-wide constraints go in AGENTS.md or CONTRIBUTING.md, reusable workflows go in the relevant .agents/skills entry, and build or release procedures go under docs. Behavioral defects are handled as code fixes with regression test proposals instead.

Does the retrospective skill apply changes automatically?

No. By default it returns findings and proposed edits only. Documentation changes are applied only when explicitly requested, and it never creates issues, sends messages, or writes memory without explicit user instruction.

What are the limitations of session retrospectives?

Retrospectives only capture lessons supported by concrete evidence from the session, so unsupported opinions and one-off preferences are excluded. They also do not replace code fixes, since behavioral defects require actual changes and regression tests.