housekeep
Skill1 starsChanged 5 months agoApache-2.0View on GitHub
Weekly hygiene pass over Iantha's memory, skills, git state, and (optional) vault. Flags stale tasks, drift in memory file timestamps, uncommitted changes, vault `_to-delete/` clutter, and skills with bloated "Learned from runs" sections. Use this whenever the user types /housekeep, mentions cleaning up / "things feel cluttered" / "let's tidy up", asks for a weekly check-in, or it's been a week since the last housekeep commit.
2 other projects use a version of this.
Adopted by
Most recently taken up by
- Dammyjay93/housekeeptheir own version, first seen Sep 26, 2026
- optave/ops-codegraph-tooltheir own version, first seen Aug 19, 2026
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- optave/ops-codegraph-tool★ 94same name
- Dammyjay93/housekeep★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 4 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 38
If the live memory files now hold real content but first-run scaffolding is still present, offer to graduate it off — the scaffolding loads …
- Very specific: line 47
Run git status and git log -5 --oneline.
- Very specific: line 51
Call /obsidian lint and surface the findings.
- Distinctive: line 67
Don't auto-archive. The user owns their tasks and vault.
- Distinctive: line 93
Default cadence: weekly. The skill is non-destructive — every change is proposed, not applied.
- Distinctive: line 95
If everything is clean: "All clear.
≈ “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.
Files
/housekeep — Weekly Memory & Skill Audit
Slim weekly hygiene pass. Scoped to Iantha's own memory, skills, git, and (if configured) vault. Single-user — no agent inboxes, no cross-project coordination.
The scan parts (memory drift, learned-section overflow, stale tasks) are bundled in scripts/audit.py so each invocation runs the same recipe deterministically. The proposal-and-apply parts stay LLM-driven because they need user judgment.
When to invoke
User typed /housekeep, OR another skill (e.g., a CLAUDE.md rule) explicitly delegates to it.
This skill does not auto-trigger from inside itself — Iantha can suggest it during /morning or /evening if it's been a week since the last housekeep commit, but the user always initiates.
Steps
1. Run the audit script
python3 .claude/skills/housekeep/scripts/audit.pyThe script auto-detects the repo root (walks up from itself looking for memory/ + .claude/). Pass --repo-root <path> to override.
This produces a report covering:
- Memory drift: files under
memory/whose*Updated:*header is older than 30 days (excludinglearnings.mdandfeedback.md, which are intentionally low-churn). Files without an*Updated:*header are skipped — they're treated as never-touched templates. - Skills with bloated learned sections: skills whose
## Learned from runssection has more than 5 entries. Suggests/consolidate-learning <skill>. - Stale tasks: items in
memory/tasks.mdwith a date older than 30 days that haven't been checked off.
Surface the report to the user.
1b. First-run graduation check (LLM-driven)
If the live memory files now hold real content but first-run scaffolding is still present, offer to graduate it off — the scaffolding loads every session yet only applies to a blank clone.
- Detect: the live
memory/*.mdhave real content (not just templates) and any of these still exist:memory/*.md.exampleseeds, a## First-Run Notessection inCLAUDE.md, or the first-run demo banner inmemory/MEMORY.md. - If both, offer once: "You're well past first-run — want me to clear out the setup scaffolding? (sample seeds + the demo notes that only apply to a fresh clone)"
- On yes, run the same graduation as
/setupstep 6:git rm memory/*.md.example, drop the MEMORY.md banner, remove the CLAUDE.md## First-Run Notessection + themorning/SKILL.mdfresh-clone demo paragraph — one commit (chore: graduate off first-run scaffolding). Keep the/setupskill (re-runnable for gap-fill). - Skip silently if memory is still blank (genuine fresh clone) or the scaffolding is already gone.
2. Git state (LLM-driven)
Run git status and git log -5 --oneline. If memory files have uncommitted changes, propose a commit. If the working tree has untracked files in memory/ or .claude/, ask the user before committing them.
3. Vault hygiene (only if vault_dir configured)
- Call
/obsidian lintand surface the findings. - Review
${vault_dir}/_to-delete/. Flag items older than 14 days for human deletion (the lint script also covers this — don't double-report).
4. Reading list (only if vault_dir configured)
Open ${vault_dir}/reading-list.md. Flag items in "Action Required" older than 14 days — surface to the user, ask if any should be dropped or escalated.
5. Propose actions, apply with confirmation
For each finding from steps 1–4, propose what to do. The user approves per item. Apply only what they confirm. Commit per category:
housekeep: archive 3 stale tasks
housekeep: vault — sweep _to-delete (5 items)Don't auto-archive. The user owns their tasks and vault.
Example report (from audit.py)
## Housekeep audit — 2026-05-01
### Memory drift (>30d unchanged, excluding frozen files) (2)
- personal.md: last updated 2026-03-25 (37d ago)
- priorities.md: last updated 2026-03-30 (32d ago)
### Skills with bloated learned sections (>5 entries) (1)
- obsidian: 7 entries
### Tasks with no movement in 30+ days (1)
- [ ] [2026-03-12] Renew car registration (date in line: 2026-03-12)
Total findings: 4Out of scope
This skill is deliberately thin. Iantha is single-user and personal — no agent inboxes, GitHub issues, market checks, cost audits, or multi-project coordination. If you need any of that, look at Iris.
Notes
- Default cadence: weekly. The skill is non-destructive — every change is proposed, not applied.
- Skip vault steps silently if
vault_diris unset. - If everything is clean: "All clear. Last housekeep YYYY-MM-DD."
Learned from runs
(empty — populated as the skill is used)
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
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.

