agentleFS
chacosoldierchacosoldier/compabob/.claude/skills/memory-debt/SKILL.md

memory-debt

Skill34 starsChanged 22 days agoMITView on GitHub

Review and apply memory updates that /reflect proposed but nobody wrote yet (the unchecked items under Suggested Memory Updates in vault/Daily/). Use /memory-debt to see what is pending, /memory-debt apply to resolve it item by item. It does not do the reflecting; that is /reflect.

The file

We split this skill into the rules it gives an agent

11 rules. About 0 are common, 6 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

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

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

    No args or review: Review Mode (show pending items).

    line 15
  2. Very specific:

    If they exist, vault/Reflections/*.md and vault/Journal/*.md (conventions some users bring from other setups).

    line 23
  3. Very specific:

    Read the target file mentioned in the proposal.

    line 32
  4. Distinctive:

    Search for the proposed content (exact match first, then semantic, keywords from the proposal).

    line 33
  5. Distinctive:

    Ask: Apply / Skip / Edit / Obsolete.

    line 64
  6. Distinctive:

    Read the target file.

    line 69
  7. Very specific:

    If the target file is new and lives at memory/topics/<slug>.md, also add a one-line pointer to memory/MEMORY.md so the index stays current.

    line 71
  8. Very specific:

    Update the diary entry from - [ ] to - [x].

    line 72
  9. Distinctive:

    Never auto-apply: Always present each item for approval before modifying any file.

    line 80
  10. Distinctive:

    Read before writing: If the content already exists, mark APPLIED silently.

    line 81
  11. Distinctive:

    Diary edits are minimal: Only flip checkboxes and update Memory Debt table status.

    line 83

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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 Debt Resolution

You are a memory hygiene tool. Your job is to find suggested memory updates that earlier sessions proposed (typically in vault/Daily/<date>.md under a ## Reflection heading, written by /reflect), verify they are still valid, and apply them with the user's approval.

Pairs with /reflect: /reflect proposes memory writes; /memory-debt collects the ones that never got applied and closes the loop.

Mode detection

Check args:

  • No args or review: Review Mode (show pending items).
  • apply: Apply Mode (resolve all pending items interactively).

Review Mode (default)

1. Scan reflection sources

  1. vault/Daily/*.md: every file with a ## Suggested Memory Updates section. This is where /reflect records its proposals.
  2. If they exist, vault/Reflections/*.md and vault/Journal/*.md (conventions some users bring from other setups).

For each file, extract:

  • "Suggested Memory Updates" entries (each starts with - [ ] or - [x]).
  • Free-text proposals under a ## Memory or ## Memory Debt heading.

2. Classify each unchecked item

For every - [ ] entry:

  1. Read the target file mentioned in the proposal. If no target is named, default to memory/topics/<slug>.md derived from the proposal title.
  2. Search for the proposed content (exact match first, then semantic, keywords from the proposal).
  3. Classify as:
    • APPLIED: content already exists in the target file. Tick the diary entry to - [x] silently.
    • PENDING: content not found, proposal is < 3 days old.
    • STALE: content not found, proposal is 3+ days old.
    • OBSOLETE: target file does not exist, or the project state contradicts the proposal.

3. Present summary

## Memory Debt Report

| # | Date | Target | Suggestion | Status | Age |
|---|---|---|---|---|---|
| 1 | YYYY-MM-DD | memory/topics/<slug>.md | <one-line summary> | PENDING | 1d |

**Totals**: X pending, Y stale, Z already applied (silent fixes), W obsolete

Apply Mode (/memory-debt apply)

1. Run Review Mode first

Build the full pending/stale/obsolete list.

2. Interactive resolution

For each PENDING or STALE item (oldest first), present:

  1. The original suggestion (from the diary).
  2. The target file's current relevant section (so the user sees what changes).
  3. The exact text to add or modify.

Ask: Apply / Skip / Edit / Obsolete.

3. Execute approved changes

For each approved item:

  1. Read the target file.
  2. Apply the change (Edit or append).
  3. If the target file is new and lives at memory/topics/<slug>.md, also add a one-line pointer to memory/MEMORY.md so the index stays current.
  4. Update the diary entry from - [ ] to - [x].

4. Report

Final summary: X applied, Y skipped, Z marked obsolete, W silent fixes.

Constraints

  • Never auto-apply. Always present each item for approval before modifying any file.
  • Read before writing. If the content already exists, mark APPLIED silently. Never duplicate.
  • MEMORY.md size cap. MEMORY.md loads every session, so keep it under about 12KB. If writing the proposal would push it past that, suggest a topic file in memory/topics/ and a one-line index pointer instead of inlining.
  • Diary edits are minimal. Only flip checkboxes and update Memory Debt table status. Never rewrite proposal text.
  • Oldest first. Process chronologically so the longest-overdue debt clears first.

Quality checklist

  • All reflection sources scanned (vault/Daily/, vault/Reflections/, vault/Journal/).
  • Each proposal's target file is actually read before classifying.
  • No duplicate content is written.
  • Diary checkboxes are updated after each application.
  • memory/MEMORY.md stays under any configured line cap.

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.