How often should I run the memory review?

It is recommended to run /si:review after completing major features, when you notice Claude repeating mistakes, or whenever your memory health report shows high line counts.

Why is the 200-line limit important in Claude Code?

Claude Code has a startup limit where it only reads the first 200 lines of memory. The /si:review command ensures your most important learned patterns stay within this limit by identifying what can be archived or promoted.

What is a 'Promotion Candidate' in auto-memory?

A promotion candidate is a pattern or rule Claude has learned through multiple sessions that is broadly useful enough to be moved from temporary memory into your permanent CLAUDE.md or rules directory.

Does the /si:review command automatically delete my data?

No, the skill generates a detailed report and actionable recommendations. You maintain control over which entries to consolidate, delete, or promote to permanent rules.

Memory Health & Review

Name: Memory Health & Review
Author: oabdelmaksoud

byoabdelmaksoud

0•

Productivity & Workflow

Analyzes Claude Code's auto-memory to promote valuable patterns to permanent rules and clean up stale or redundant information.

The Review skill is an essential utility for maintaining a high-performance development environment by auditing Claude's auto-memory. It identifies recurring patterns that should be promoted to permanent project rules in CLAUDE.md, detects stale entries referencing non-existent files, and flags contradictions between memory and existing instructions. By keeping your memory files concise and relevant, this skill prevents context bloat and ensures Claude's self-improving capabilities remain accurate and efficient.

Key Features

01Identification of 'Promotion Candidates' for permanent project rules

02Conflict checking against existing CLAUDE.md and .claude/rules/

030 GitHub stars

04Comprehensive audit of MEMORY.md and specialized topic files

05Memory health reporting including line counts and startup limit tracking

06Stale entry detection for outdated commands or deleted files

Use Cases

01Optimizing agent performance when MEMORY.md exceeds the 200-line startup limit

02Cleaning up project context after major refactors or file deletions

03Formalizing recurring feedback into permanent coding standards

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add oabdelmaksoud/agention review

For use in Claude.ai and ChatGPT