madness

Automate AI collaboration retrospectives with init, mid-review, and final modes.

Updated Feb 23, 2026
One-click install
npx skills add https://github.com/Gary-Ge-xixi/madness --skill madness
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: madness
Source: https://github.com/Gary-Ge-xixi/madness/tree/main
Command: npx skills add https://github.com/Gary-Ge-xixi/madness --skill madness

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires pypdf, pdfplumber, pdf2image, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill provides a structured retrospective workflow for AI collaboration, extracting actionable improvements from session records to create a bidirectional feedback loop. It transforms past reviews into verifiable, evolving rule assets, continuously optimizing human-AI collaboration quality.

Core Features & Use Cases

  • Structured Retrospectives: Offers three modes (init, mid, final) for comprehensive project review.
  • Asset System: Manages "Genes" (methodology), SOPs (procedures), and Prefs (preferences) with a confidence lifecycle.
  • Socratic Questioning: Engages users with targeted questions to uncover cognitive blind spots and refine rules.
  • CLAUDE.md Injection: Automatically updates AI instructions via a "Reflection" mechanism for continuous learning.
  • Cross-Team Sharing: Generates portable.json for sharing refined methodologies across teams.

Quick Start

Use the madness skill to perform a project-level AI collaboration retrospective.

Frequently Asked Questions about madness

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

FAQPage Schema
How do I automate AI collaboration retrospectives to extract actionable improvements?

Automating AI collaboration retrospectives involves extracting actionable improvements from session records to create a bidirectional feedback loop, transforming past reviews into verifiable, evolving rule assets.

What is the best way to update CLAUDE.md rules based on project review feedback?

Updating CLAUDE.md rules is achieved through a Reflection mechanism that automatically injects refined instructions, ensuring AI collaboration quality continuously improves from retrospective insights.

How does Socratic questioning identify cognitive blind spots during AI collaboration reviews?

Socratic questioning identifies cognitive blind spots by engaging users with targeted questions during retrospectives, systematically uncovering hidden assumptions to refine collaboration rules and methodology.

Can I share refined AI collaboration methodologies across multiple project teams?

You can share refined methodologies across teams by generating a portable.json file, which packages the extracted Genes, SOPs, and preferences for cross-team distribution and reuse.

What are the three modes for conducting a project-level AI collaboration retrospective?

The three modes for project-level retrospectives are init for baseline analysis, mid-review for diagnosis and learning, and final review for extracting cross-project methodology improvements.

Do I need PDF processing dependencies to run retrospective analysis on session records?

You need pypdf, pdfplumber, and pdf2image dependencies to process session records, as these libraries handle the PDF parsing required to extract actionable insights for the retrospective workflow.