shine-retro

Analyze Claude learning logs and propose memory updates in markdown.

1|Updated Apr 15, 2026
One-click install
npx skills add https://github.com/diShine-digital-agency/SHINE-Code-System --skill shine-retro
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: shine-retro
Source: https://github.com/diShine-digital-agency/SHINE-Code-System/tree/main/skills/shine-retro
Command: npx skills add https://github.com/diShine-digital-agency/SHINE-Code-System --skill shine-retro

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Cross-session retrospectives on Claude memory are time-consuming and error-prone; this skill reads learning logs and memory placeholders and suggests concrete updates to typed memory to improve future sessions.

Core Features & Use Cases

  • Read learning logs: review last N days of per-turn data and per-session blocks.
  • Propose memory updates: generate targeted memory additions or adjustments without overwriting existing memory.
  • Open loop governance: present a safe path for applying updates only with explicit approval or --apply.

Quick Start

Run shine-retro with an optional --days N to review the last N days of learning logs and, if you want the changes applied, use --apply to create an external retro memory file.

Frequently Asked Questions about shine-retro

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

FAQPage Schema
How do I generate memory update proposals from Claude learning logs?

To generate memory update proposals from Claude learning logs, you can run a cross-session retrospective that analyzes the metadata in your learning-log.jsonl and learning-log.md files. This process identifies patterns and suggests concrete memory additions without overwriting existing memory.

What is the best way to review cross-session memory retro without overwriting existing data?

Reviewing cross-session memory retro safely requires a process that only proposes updates without modifying existing memory. You can analyze learning logs to generate a concise markdown retrospective with activity snapshots, gaps, proposed updates, and open questions.

Can I apply proposed memory updates automatically using learning-log metadata?

You can apply proposed memory updates by using the explicit --apply flag. This creates an external retro memory file based on analyzed learning-log metadata, ensuring no existing memory is overwritten without explicit approval.

How do I limit a retrospective analysis to the last N days of learning logs?

To limit retrospective analysis to a specific timeframe, run your memory retro tool with the --days N parameter. This restricts the pattern identification and memory-update proposal generation to only the learning log data from the last N days.

Does Claude memory management support external retro files for learning logs?

Claude memory management can support external retro files when you explicitly approve updates via the --apply flag. This creates a separate file for proposed memory additions, ensuring your original typed memory remains unmodified during standard analysis.