agentleFS
fortunto2fortunto2/solo-factory/skills/memory-audit/SKILL.md

solo-memory-audit

Skill18 starsChanged 4 months agoMITView on GitHub

Use when "audit memory", "check CLAUDE.md", "memory map", "rules audit", "context budget", "what loads in my session", or need to optimize agent context loading.

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 3 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    If all argument, add --all-projects flag instead of path.

    line 43
  2. Very specific:

    If uv is not available, fall back to:

    line 45
  3. Distinctive:

    Analyze the output: Read the tree display and audit hints table.

    line 50
  4. Distinctive:

    Verify parent CLAUDE.md files exist (inheritance chain intact)

    line 70
  5. Distinctive:

    Check that project CLAUDE.md has essential sections (structure, commands, stack)

    line 71
  6. Very specific:

    Verify .claude/rules/ files have valid paths: frontmatter syntax

    line 72

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “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-audit

Audit Claude Code memory hierarchy for a project. Shows what files load at session start, total context budget, and optimization hints.

Context

Claude Code loads memory files at session start in this order:

  1. Managed policy (/Library/Application Support/ClaudeCode/CLAUDE.md)
  2. User memory (~/.claude/CLAUDE.md)
  3. User rules (~/.claude/rules/*.md)
  4. Auto-memory (~/.claude/projects/{key}/memory/MEMORY.md, first 200 lines)
  5. Project hierarchy (walk from / to CWD: CLAUDE.md, .claude/CLAUDE.md, .claude/rules/*.md)
  6. Local overrides (CLAUDE.local.md at each level)

Rules with paths: frontmatter are conditional — loaded only when working on matching files. Rules without paths: always load and consume context budget.

Target: keep startup context under 40k chars. Large CLAUDE.md files should extract domain-specific sections into conditional .claude/rules/*.md files.

Steps

  1. Determine target. Parse $ARGUMENTS:
    • Empty or . -> use CWD
    • Path -> use that path
    • all -> scan CWD subdirectories that have CLAUDE.md
  2. Run memory_map.py. Execute:

    uv run python ${CLAUDE_PLUGIN_ROOT}/scripts/memory_map.py <path> --audit

    If all argument, add --all-projects flag instead of path.

    If uv is not available, fall back to:

    python ${CLAUDE_PLUGIN_ROOT}/scripts/memory_map.py <path> --audit --plain
  3. Analyze the output. Read the tree display and audit hints table. Key metrics:
    • Total startup chars — should be under 40k
    • Large files (>300 lines) — candidates for splitting into rules
    • Unconditional rules (>30 lines, no paths: frontmatter) — add paths: to make conditional
    • Duplicate sections — same ## Header in multiple files, extract to one place
    • Missing auto-memory — project doesn't learn across sessions
  4. Generate recommendations. Based on audit hints, suggest concrete actions:

    IssueAction
    File > 300 linesExtract domain sections to .claude/rules/{topic}.md with paths: frontmatter
    Total > 40k charsIdentify largest files, move conditional content to rules
    Unconditional rule > 30 linesAdd paths: frontmatter targeting relevant source files
    Duplicate sectionsKeep in the highest-level file, remove from children
    No auto-memoryNormal for new projects, just informational
    User-level rule is project-specificMove from ~/.claude/rules/ to project's .claude/rules/
    Generic rule in projectMove up to ~/.claude/rules/ or parent CLAUDE.md
  5. Check hierarchy health. For the target project:
    • Verify parent CLAUDE.md files exist (inheritance chain intact)
    • Check that project CLAUDE.md has essential sections (structure, commands, stack)
    • Verify .claude/rules/ files have valid paths: frontmatter syntax
  6. Output report:
    ## Memory Audit Report
    
    **Project:** [path]
    **Startup context:** X files, ~Y chars (Z% of 40k budget)
    
    ### Tree
    [paste tree output from memory_map.py]
    
    ### Issues Found
    [from audit hints]
    
    ### Recommendations
    1. [specific action with file path]
    2. [specific action with file path]
    
    ### Hierarchy Health
    - Inheritance chain: [OK / broken at level X]
    - Auto-memory: [present / missing]
    - Rules: X total (Y conditional, Z always-on)

Rules Frontmatter Reference

Conditional rules load only when the agent works on files matching the glob pattern:

---
paths:
  - "src/lib/analytics/**"
  - "src/app/api/analytics/**"
---
# Analytics Module
...content only loaded when working on analytics files...

Without paths:, the rule always loads and consumes context budget.

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.