agentleFS
neskuchnyneskuchny/content_factory/.claude/skills/memory-compaction/SKILL.md

memory-compaction

Skill1 starsChanged 2 days agoMITView on GitHub

Compact the client's memory: a week folds into journal lines, a quarter into a summary, and MEMORY.md stays a short index. Use at the retro at the end of the week and once a month.

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 1 is distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

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

    Update memory/MEMORY.md as an index (≤ 200 lines):

    line 14
  2. Distinctive:

    Check for repeats: topics that came out ≥ 3 times in 2 months → "on pause".

    line 26

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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 compaction

Goal: at the start of work the agent reads little but knows the essentials: what has already happened, what works, where we are now.

Weekly (after the retro)

  1. For each published unit, one line in memory/published-log.md: | 2026-10-21 | W43-R02 | instagram | Site walkthrough | finished install on site, "no drafts" | pain_direct | reels/f01 | 2.4 | WIN |
  2. Full scripts and files stay in weeks/<W>/ (this is the archive; it is not read by default).
  3. Update memory/MEMORY.md as an index (≤ 200 lines):
    • state (onboarding, current week, open questions);
    • the 3–5 main conclusions of recent weeks with links to files;
    • topics on pause (overheated) and candidate topics;
    • the client's latest feedback in one line.

Monthly

  1. Fold published-log lines older than a quarter into a "Summaries" block: "Q3: 12 Reels about installation; best performer: before/after with a number in the first line (median ×1.9); the 'prices' topic is overheated — paused until December".
  2. In targeting-map, hooks-library and platform-learnings: delete conclusions refuted by fresh data (with a note in changelog.md).
  3. Check for repeats: topics that came out ≥ 3 times in 2 months → "on pause".

Where to remember what (short)

WhatWhereChanged by
Business, product, taboos, permissionspassport.yaml, product-base/, permissions.yaml, CLAUDE.mdhuman only (locked layer)
Style and tone, learned rulesbrand/voice.md, brand/visual.md, brand/tokens.jsonfeedback, with evidence
What was publishedmemory/published-log.mdretro
What workstargeting-map, hooks-library, platform-learnings, channel-stateanalyst
Numbersanalytics/*.jsonlscripts only
Golden references and assetsbrand/golden/, brand/assets/, Drivefeedback, human
An agent's experience in its own craft.claude/agent-memory/<agent>/MEMORY.mdthe agent itself (not about the client)

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.