agentleFS
HurricaHjzHurricaHjz/SecondSelf/.claude/skills/weekly-audit/SKILL.md

weekly-audit

Skill150 starsChanged 8 days agoMITView on GitHub

The weekly review's tools: the fresh-eyes audit (AUDIT.md), the waste report (weekly-waste.py) and the rule register's one pass (rule_register.py). The calendar's Weekly review runs them each Sunday; /weekly-audit runs them on request. Nothing here schedules itself or stops other work.

  • Installs packages

6 other projects use a version of this.

Adopted by

Most recently taken up by

The file

Files
SKILL.md

Weekly audit

The Weekly review in CALENDAR.md runs these tools, in this order; nothing schedules them, and each tool's procedure is its own file.

ToolWhat it doesIts procedure
The fresh-eyes auditOne reviewer with no prior context reads the week's changed framework files, the lifecycle rows due for review, the items with no row and one rotating slice; the head re-verifies and disposes of each finding.AUDIT.md, with slice.py, pass_order.py and brief-audit.md
The rule register's one passrule_register.py weekly lists what the rule register finds, inside the audit (AUDIT.md, step 8). It only lists.rule_register.py's own docstring
The clean-up's listingrule_register.py cleanup prints the store sizes and the old records no wiki page, calendar line or open register entry names, for the weekly review's clean-up step. It lists and deletes nothing.rule_register.py's own docstring
The waste reportRecorded helper consumption and operational outcomes across runs, from the run records alone. It judges nothing.README.md and FIELDS.md, with weekly-waste.py

Where things live

This folder ships, so it holds code and procedure only. Everything a run produces stays out of it:

  • State, outside the vault: <store>/tools/weekly-audit/slices.json and <store>/tools/weekly-waste/dispositions.jsonl, where <store> is LLM_WIKI_STORE, else ~/.llm-wiki. Never delete either to start over: each holds a history the next run depends on.
  • Reports, in the user's layer: output/reports/audit-report-<date>.md, output/reports/waste-report-<date>.md and output/reports/waste-detail-<date>.md. The shipped .gitignore keeps output/ out of every push.
  • Rows: wiki/developments/check-lifecycle.md, which setup.sh seeds with its rules and no rows.
  • The rule register's data: wiki/developments/rule-register-data.json. It holds the owner's wordings and never ships, so on a fresh copy the register's pass says "not applicable".

First use

setup.sh makes both stores on a fresh vault. By hand, once each:

store="${LLM_WIKI_STORE:-$HOME/.llm-wiki}"
mkdir -p "$store/tools/weekly-audit" "$store/tools/weekly-waste" "$store/hooks" output/reports
python3 -B .claude/skills/weekly-audit/slice.py --init
python3 -B .claude/skills/weekly-audit/weekly-waste.py init

The audit's slice.py needs markdown-it-py 4.2.0 (python3 -m pip install markdown-it-py==4.2.0); without it every pick stays incomplete and says so. The waste report and pass_order.py need nothing beyond Python itself.

The suites are test_slice.py, test_pass_order.py, test_weekly_waste.py and test_rule_register.sh; each ends PASS n/n or FAIL k/n.

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.