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 hams-ollo/zen-agent-skills --skill house-reviewgit clone --depth 1 https://github.com/hams-ollo/zen-agent-skillsWrote 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/hams-ollo/zen-agent-skills/house-review)<a href="https://agentmods.dev/skills/hams-ollo/zen-agent-skills/house-review"><img src="https://agentmods.dev/badge/skills/hams-ollo/zen-agent-skills/house-review/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/hams-ollo/zen-agent-skills/house-review"><img src="https://agentmods.dev/badge/skills/hams-ollo/zen-agent-skills/house-review.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 | $0.00180 | $0.03589 |
| Opus 5 | $0.00090 | $0.01795 |
| Sonnet 5 | $0.00036 | $0.00718 |
| Haiku 4.5 | $0.00018 | $0.00359 |
Grade A, and why
house-review 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 7d 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 — 243 lines — stays where its author put it; the contents beside it link to each section on GitHub.
house-review
A house-style code review with an explicit rubric and severities. It reads a change, applies the
review-quality lens, and writes a structured review. It is
report-only: it never edits or commits. Reconciling the findings is the author's call, handled
by the human, /simplify, or fix-batch.
It is named for the house rubric it applies rather than for the act of reviewing, which also keeps it clear of the review commands several harnesses ship built in: a skill cannot resolve a namespace collision by asserting it is different, so the name avoids one instead. It composes a swappable lens (the moonray pattern: a reusable review "shot" a workflow composes, with findings validated before they are reported), so the review bar lives in one editable file and future quality skills can reuse it.
What it produces
A severity-ranked markdown review: a one-line verdict, then findings ordered blocker to nit,
each with a file:line anchor, an exact quote of the cited code, a stable signature, the issue, why
it matters, and a concrete suggested fix. When nothing substantive survives validation, it says so
plainly.
Design choices
Settled decisions (resolved with the user); these are not up for re-litigation:
- Rubric and severities live in the lens,
review-quality.md: eight categories, and theblocker/major/minor/nitscheme. Edit the lens to retune the bar; it is swappable likehouse-style.md. - Report-only. The skill never edits or commits. Each finding carries a concrete fix in text.
- Findings are validated before reporting (the lens's govern/revalidate step). A confident false positive costs more than a missed nit.
- Validation is checked, not trusted. Every finding's quoted evidence is resolved against the file before the finding is reported, and one that resolves nowhere is dropped rather than hedged. The rule and its branches live in the lens; Step 4 below applies them.
- Every reported finding carries a signature, defined by the lens, so the same defect found by two reviewers or across three runs is one countable defect. This skill emits it and reads nothing from it.
- Three review modes: an explicit path scope reviews named files in full; an explicit base or
range reviews that range as given, winning over the default and reported as supplied rather than
resolved; otherwise the default is the branch vs its merge-base with the default branch (with a
working-tree fallback), reusing
pr-describe's changeset logic. A path scope and a range together narrow the second mode rather than producing a fourth.
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.
- 7d ago Changed · +7 lines cc833cc340ce
- 11d ago First seen · 236 lines · 180 tokens per session scan A 4638d570a5ad
house-review is a skill published in the GitHub repository hams-ollo/zen-agent-skills (2 stars, last pushed 10d ago), licensed MIT. It adds 180 tokens to every session and 3,589 once invoked, about $0.0009 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
backlog-reconcile
Use when auditing, reconciling, or verifying a repo's meta/BACKLOG.md Active queue against reality, "audit the backlog", "reconcile the board", "is the board still true", "dọn backlog", "rà soát backlog", "check board rows against their specs", or when a scheduled run asks for the backlog-reconcile audit of a repo.…
setup-rootloom
Plan, install, inspect, update, or roll back Rootloom Personal Core in a user's Codex home. Supports Skills-only, guidance, and the recommended personal preset. Use when the user explicitly asks for a Rootloom setup plan, installation, configuration, bootstrap, repair, audit, status, update, reduction, rollback, or…
project-memory
Initialize, retrieve, and explicitly update the separate Rootloom Memory plugin's repository-owned architecture, risk, decision, and failure lessons. Use only when explicitly requested. Memory is experimental, advisory, and outside Rootloom Core.
project-guidance
Create, refine, or validate repository guidance on explicit user request; preserve real-path scope, managed ownership, and unrelated content.
doc-drift
Use for the whole-estate doc audit, "run the doc-drift loop", "audit the docs against the code", "are the docs still true", "doc drift sweep", "check every doc claim", or a scheduled doc-audit cadence run. Enumerates every LIVING doc (README, MANUAL, AGENTS, WORKFLOW, architecture, patterns), verdicts each against the…
gauntlet-proof-audit
Use to audit the gauntlet records, "are the ROUNDS.md records honest", "verify the proof corpus", "check each run record against its evidence", "audit the gauntlet corpus", "is the gauntlet proof trustworthy", or a scheduled gauntlet-proof-audit cadence run. Enumerates every committed gauntlet run record…