Can I run this on retrospectives that were already processed?

Yes, you can use the --reprocess flag to force the skill to re-extract and update items from an existing retrospective file.

How does the skill handle duplicate knowledge items?

It performs semantic similarity checks against existing items in your knowledge directory. If a match is found, it can either supersede the old item or cross-reference them via tags.

What types of files can this skill process?

The Retro skill is designed to process Markdown retrospective files located in the .swarm/retrospectives/ directory.

What is 'Scout Hoisting'?

Hoisting is the process of marking specific knowledge items as high-value candidates to be moved or 'hoisted' from a project-specific level to a broader organizational or 'Scout' level.

What are the knowledge dimensions like Epistemology and Qualia?

These are classification frameworks used to organize knowledge: Epistemology focuses on decisions and facts, Qualia on subjective experiences/improvements, and Praxeology on successful actions.

Retro Knowledge Extractor

Name: Retro Knowledge Extractor
Author: hgeldenhuys

byhgeldenhuys

0•

学習とドキュメンテーション

Extracts, classifies, and deduplicates institutional knowledge from project retrospectives to build a structured, persistent knowledge base.

The Retro skill automates the critical task of transforming team retrospectives into actionable, structured knowledge. By parsing standard retrospective sections like 'What Went Well' and 'Key Decisions,' it identifies and classifies insights into Epistemology, Qualia, and Praxeology dimensions. The skill features built-in semantic deduplication to keep project documentation clean and includes 'scout hoisting' logic to flag high-value insights for organizational consolidation, ensuring that valuable lessons learned during development are never lost.

主な機能

01Identification of 'hoistable' items for high-level consolidation

02Dimensional classification (E/Q/P) for structured learning

03Automatic frontmatter updates and cross-referencing

04Semantic deduplication against existing knowledge items

050 GitHub stars

06Automated knowledge extraction from Markdown retrospectives

ユースケース

01Building a searchable knowledge base from project post-mortems

02Maintaining technical standards through automated decision-tracking

03Identifying high-value learnings for team-wide or organizational scaling

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add hgeldenhuys/recursive-ai retro

For use in Claude.ai and ChatGPT

Download Skill