Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add OKHP3/skillz --skill okhp3-repository-janitorgit clone --depth 1 https://github.com/OKHP3/skillzWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/okhp3/skillz/okhp3-repository-janitor)<a href="https://agentmods.dev/skills/okhp3/skillz/okhp3-repository-janitor"><img src="https://agentmods.dev/badge/skills/okhp3/skillz/okhp3-repository-janitor/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/okhp3/skillz/okhp3-repository-janitor"><img src="https://agentmods.dev/badge/skills/okhp3/skillz/okhp3-repository-janitor.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00099 | $0.01236 |
| Opus 5 | $0.00049 | $0.00618 |
| Sonnet 5 | $0.00020 | $0.00247 |
| Haiku 4.5 | $0.00010 | $0.00124 |
Grade A, and why
okhp3-repository-janitor scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 11d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 81 lines — stays where its author put it; the contents beside it link to each section on GitHub.
okhp3-repository-janitor
OverKill Hill P³ · overkillhill.com · github.com/OKHP3
Keep a multi-repository Git estate understandable without losing work. Treat origin/main as the comparison baseline, not as evidence that other work is disposable.
Safety contract
- Begin with a read-only inventory. Do not delete repositories, branches, stashes, refs, or commits during discovery.
- Preserve before pruning. Record dirty files, stashes, local-only commits, and unreachable commits. Pin unreachable commits under a dated
refs/archive/namespace before a later cleanup can trigger garbage collection. - Refresh remotes with
git fetch origin; do not use--pruneuntil stale remote branches have been classified. - Compare each checkout to the refreshed
origin/main. Report both file-level and commit-level differences. - Treat a merged branch, a closed pull request, and an abandoned branch as different facts. Query the pull request before deletion.
- Merge only a reviewed pull request whose checks and destination are suitable. Do not merge a branch merely because it is old or has a bot name.
- Delete a remote branch only when its merged/superseded status, commit reachability, and pull-request state are confirmed. Delete the local tracking branch only after the remote deletion is verified.
- Never rewrite
main, force-push, or remove stashes, archive refs, or untracked files without a separately stated and confirmed recovery plan.
Audit workflow
- Resolve the mirror root and verify that every child is a Git checkout. Keep repositories outside that root out of scope.
- Run
scripts/audit_mirrors.py <mirror-root> --include-unreachablefor a read-only baseline. Add--fetchto refresh remote-tracking refs without changing any working tree. - For each repository, inspect the report sections in this order:
- dirty working-tree files and stashes;
- commits reachable locally but not from a remote;
- current
HEADversusorigin/main, distinguishing its direct tree difference from the files changed on the branch since the shared base; - local and remote branches and whether each is already reachable from
origin/main; - archive refs and unreachable commits.
- Query GitHub for every non-main remote branch: its pull request, whether it is open, closed, or merged, its checks, and whether a newer branch supersedes it. Read
references/branch-lifecycle.mdbefore classifying candidates. - Produce a decision ledger with one row per candidate:
keep,review,merge,close PR,archive, ordelete. State the evidence and recovery point. - Execute in small batches. For each approved merge, refresh, verify the expected head, merge through the pull request, refetch, then delete only the exact verified merged branch. Re-run the audit after every batch.
What ships with it
3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 11d ago First seen · 81 lines · 99 tokens per session scan A 5a5d1a14705f
okhp3-repository-janitor is a skill published in the GitHub repository OKHP3/skillz (3 stars, last pushed today), licensed MIT. It adds 99 tokens to every session and 1,236 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
release-skills
A release workflow for preparing and publishing a new version of a software project. A release is a named version that may include updated version files, release notes, tags, or a GitHub Release.
pm-release
A release-planning workflow for preparing software to go live, including checks, deployment choices, timing, and rollback conditions.
swe-workflow
Orchestrates the full five-stage flow from raw idea to shipped PR — grill-with-docs → to-prd → to-issues → triage → worktree+planning-with-files. Each stage answers one question (What do I want? / What does done look like? / What are the units of work? / What's actionable? / Build it). Use when the user has an idea…
private-fork-sync
Make a public GitHub fork private and keep it synced with its upstream parent. Use when the user wants to make a fork private, or to pull/sync newer commits from the upstream parent into their fork.
skill-repository-maintainer
A maintenance guide for checking, auditing, and synchronising a Codex Skill repository. It covers package structure, metadata, release boundaries, and differences between the source and installed folders.
git-worktree
Use this skill whenever a task involves running parallel work on a git repository without conflicts — e.g. "fix this bug while I keep working on my feature", "run Claude on two branches at once", "work on this task in isolation", "create a throwaway branch for this experiment", or anything where two agents or sessions…