How does it calculate the correction rate?

It analyzes your session history for similar tasks and calculates the ratio of edits to corrections to flag areas that historically required more effort.

When is the best time to use this skill?

You should trigger Replay Learnings at the very start of a new task or when moving into a sensitive area of the codebase like authentication or middleware.

Does it prioritize new or old learnings?

The skill is designed to rank learnings by relevance to your current task rather than just chronological recency, ensuring the most applicable advice comes first.

How does Replay Learnings find relevant information?

It extracts keywords from your task and performs targeted searches across memory files like .claude/LEARNED.md and CLAUDE.md to find matching patterns.

What if there are no existing logs for my task?

If no relevant learnings are found, the skill will explicitly state so and recommend starting with a 'scout' agent to explore the task area first.

Replay Learnings

Name: Replay Learnings
Author: rohitg00

byrohitg00

•

1,442

•

Productivity & Workflow

Retrieves and surfaces project-specific insights and past mistakes to prevent error repetition before starting new tasks.

Replay Learnings serves as a proactive memory layer for your development workflow, ensuring you never make the same mistake twice. By automatically extracting keywords from your current task, it scans your project's memory files and historical logs to generate a 'Replay Briefing' that ranks past learnings by relevance. This skill helps developers maintain high code quality by surfacing contextual guardrails, identifying high-risk areas based on previous correction rates, and providing actionable advice derived from actual project history.

Key Features

01Session history analysis to identify tasks with high correction rates

02Relevance-based ranking to prioritize impactful historical insights

031,442 GitHub stars

04Contextual briefings that include original mistake details for clarity

05Automated keyword extraction from task descriptions for targeted search

06Multi-source memory scanning across LEARNED.md, CLAUDE.md, and logs

Use Cases

01Reviewing previous architectural mistakes before starting a complex refactor

02Identifying high-friction code modules that previously required multiple corrections

03Recalling specific mocking patterns for authentication tests to avoid flaky builds

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add rohitg00/pro-workflow replay-learnings

For use in Claude.ai and ChatGPT

Download Skill