retro

Consolidate cross-agent memory learnings into actionable retro summaries.

4|Updated Feb 1, 2026

Search Tags:#automation #learning #patterns #insights #memory #retro

Ready to run

npx skills add https://github.com/mkalkere/agent-coordinator --skill retro-mkalkere

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: retro
Source: https://github.com/mkalkere/agent-coordinator/tree/main/.os/skills/lifecycle/retro
Command: npx skills add https://github.com/mkalkere/agent-coordinator --skill retro-mkalkere

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Cross-agent teams often end up with fragmented learnings scattered across memories and references. Retro consolidates these learnings into a single, actionable retro, reducing duplication of effort and accelerating improvements.

Core Features & Use Cases

Scan agent memories to extract learnings and patterns
Categorize findings into patterns, anti-patterns, gaps, and tool insights
Propagate high-value learnings to shared memory and update skill references

Quick Start

Run weekly retro to consolidate learnings from memory and publish a retro summary to the shared memory store.

Frequently Asked Questions about retro

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

FAQPage Schema

How do I consolidate cross-agent learnings from fragmented memory files?▼

To consolidate cross-agent learnings, you can parse and aggregate memory data from .os/agents/*/memory directories, categorize findings into patterns and anti-patterns, and write a summary to .os/retro.

What is the best way to run a weekly retro for multiple agents?▼

A weekly retro scans agent memories to extract learnings and patterns, categorizing them into gaps and tool insights, then propagates high-value learnings to shared memory via the coordinator.

Does the retro consolidation process work in standalone mode without a coordinator?▼

Yes, retro consolidation works in both framework and standalone modes, parsing memory data and writing summaries to .os/retro even when a coordinator is unavailable to propagate insights.

When should I run a cross-agent retro to extract actionable insights?▼

You should run a cross-agent retro on a weekly cadence or after major milestones to reduce duplicated effort, extract patterns from scattered memories, and accelerate improvements across agents.