agentleFS
cveralyoncveralyon/axel-setup/skills/memory-review/SKILL.md

memory-review

Skill4 starsChanged 5 months agoMITView on GitHub

Review, optimize, and maintain the persistent memory system. Reads all memory files, identifies stale/duplicate/outdated entries, asks the user targeted questions, and produces a clean, optimized memory state. Use when memory feels bloated, after major project changes, or periodically to keep context sharp.

7 other projects use a version of this.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

14 rules. 0 are common, 14 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 6 edited versions we read keep these.

  • 14

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read every memory file listed in the index

    line 15
  2. Distinctive:

    Ask the user targeted questions to resolve ambiguities:

    line 49
  3. Distinctive:

    Keep questions grouped and numbered for easy answering.

    line 55
  4. Distinctive:

    Update stale files with correct information

    line 60
  5. Distinctive:

    Remove files the user confirms as unnecessary

    line 62
  6. Distinctive:

    Add any new memory files the user requests

    line 63
  7. Distinctive:

    NEVER delete a memory file without explicit user confirmation

    line 78
  8. Distinctive:

    NEVER remove information that is still relevant — when in doubt, ask

    line 79
  9. Distinctive:

    Prefer updating over removing — information decay is better handled by correction than deletion

    line 80
  10. Distinctive:

    Keep MEMORY.md index lines under 120 chars

    line 81
  11. Distinctive:

    After cleanup, verify all index links point to existing files

    line 83
  12. Distinctive:

    After major team/project changes

    line 98
  13. Distinctive:

    When memory files exceed 20 entries

    line 99
  14. Distinctive:

    When you notice conflicting information in memory

    line 100

“Common” is measured: kept by at least 6 in 10 of the 6 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Memory Review & Optimization

Interactive memory maintenance skill. When invoked, systematically reviews all memory files and works with the user to keep the memory system clean, relevant, and efficient.

Workflow

Phase 1: Audit

  1. Read ~/.claude/memory/MEMORY.md (index)
  2. Read every memory file listed in the index
  3. Detect any orphan files (exist on disk but not in index)
  4. For each memory file, evaluate:
    • Staleness: Does the content reference things that may have changed? (team members, tools, architectures)
    • Redundancy: Is this duplicated or overlapping with another memory file?
    • Relevance: Is this still useful for future conversations?
    • Accuracy: Does this match current reality? (verify against code/git if needed)
    • Size: Is this too verbose? Can it be trimmed without losing value?

Phase 2: Report

Present a structured report to the user:

## Memory Audit Report

### Summary
- Total files: X
- Index entries: X
- Orphan files: X (not in index)
- Potentially stale: X
- Candidates for merge: X
- Candidates for removal: X

### Per-File Analysis
For each file, show:
- Status: OK / STALE / MERGE_CANDIDATE / REMOVE_CANDIDATE
- Last modified: date
- Reason: why this status
- Suggested action: keep / update / merge with X / remove

Phase 3: Interactive Q&A

Ask the user targeted questions to resolve ambiguities:

  • "Is X still accurate?"
  • "Should I merge A and B since they overlap?"
  • "This mentions Y which seems outdated — remove or update?"
  • "Any new context I should add?"

Keep questions grouped and numbered for easy answering. Max 5-7 questions per round.

Phase 4: Execute

Based on user answers:

  1. Update stale files with correct information
  2. Merge redundant files
  3. Remove files the user confirms as unnecessary
  4. Add any new memory files the user requests
  5. Rebuild MEMORY.md index with accurate one-line descriptions
  6. Sync both memory locations:
    • ~/.claude/memory/ (shared, primary)
    • ~/.claude/projects/ (legacy fallback, path varies by user)

Phase 5: Summary

Show final state:

  • Files kept / updated / merged / removed
  • Total memory size before vs after
  • Next review suggestion (based on change velocity)

Rules

  • NEVER delete a memory file without explicit user confirmation
  • NEVER remove information that is still relevant — when in doubt, ask
  • Prefer updating over removing — information decay is better handled by correction than deletion
  • Keep MEMORY.md index lines under 120 chars
  • Memory files should use the standard frontmatter format (name, description, type)
  • After cleanup, verify all index links point to existing files

Memory Types Reference

TypePurposeDecay Rate
userIdentity, preferences, skillsLow — rarely changes
feedbackHow to work with userLow — corrections persist
projectOngoing work, team, goalsHigh — changes frequently
referenceWhere to find external infoMedium — tools change

When to Suggest Running This

Suggest /memory-review to the user when:

  • More than 30 days since last review
  • After major team/project changes
  • When memory files exceed 20 entries
  • When you notice conflicting information in memory

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.