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 andr-ca/agentharness --skill audit-review-followupgit clone --depth 1 https://github.com/andr-ca/agentharnessWrote 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/andr-ca/agentharness/audit-review-followup)<a href="https://agentmods.dev/skills/andr-ca/agentharness/audit-review-followup"><img src="https://agentmods.dev/badge/skills/andr-ca/agentharness/audit-review-followup/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/andr-ca/agentharness/audit-review-followup"><img src="https://agentmods.dev/badge/skills/andr-ca/agentharness/audit-review-followup.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 | $0.00042 | $0.00971 |
| Opus 5 | $0.00021 | $0.00485 |
| Sonnet 5 | $0.00008 | $0.00194 |
| Haiku 4.5 | $0.00004 | $0.00097 |
Grade A, and why
audit-review-followup 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 12d 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 — 87 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Audit Review Follow-up
Assess whether a past review's recommendations were genuinely implemented — not just marked done — and re-score the repo. This is an assessment task: report findings, do not fix anything unless asked.
The Prompt (canonical form)
Check the review and its status report in
docs/operational/reviews/, then verify what was actually implemented against the current repo state — do not trust the status report's checkmarks. Did it close all the gaps? What gaps did the status report itself miss? What would you add next? Re-score using the original review's dimensions.
Procedure
1. Locate the documents
- Reviews live in
docs/operational/reviews/as<name>-review.md(findings + scored verdict) and<name>-review-status.md(per-item disposition). - If several review cycles exist, use the frontmatter datestamps (required per
docs/operational/README.md) to pick the cycle in question — usually the newest.
2. Read both documents fully
Extract: the item list (usually a numbered backlog), the claimed status of each item, and the original scoring rubric/dimensions.
3. Verify claims — never trust checkmarks
For each item marked done, check the repo itself. Typical checks:
- Files claimed created/deleted:
ls,test -e, read key sections. - CI claimed added/passing:
gh run list— confirm green on the default branch, not just "workflow file exists". - Repo settings claimed changed (branch protection, rename, tags):
gh api,git remote -v,git tag,git config core.hooksPath. - "Removed everywhere" claims (a phrase, a bad pattern):
grep -rnacross the repo — the sweep that was run may have missed non-link prose. - Fixed scripts: read the fix and, where cheap, execute it.
Spot-check breadth over depth: every category of claim, not every single item.
4. Hunt the missed instances (the highest-value step)
Status reports fail in classes, not one-offs. When you find one leftover, ask
what verification method produced it and where else that method is blind.
Example: a markdown link checker validates [text](path) but not prose
asserting a file exists — so grep for the claim text, not just dead links.
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.
- 12d ago First seen · 87 lines · 42 tokens per session scan A 5aefc0ac5ca2
audit-review-followup is a skill published in the GitHub repository andr-ca/agentharness (1 stars, last pushed 3d ago), licensed MIT. It adds 42 tokens to every session and 971 once invoked, about $0.0002 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
agentplane-task-closure-recovery
Use when Agentplane task completion, direct finish, branchpr integration, hosted-close, close-tail PRs, PR metadata, dirty task artifacts, or remote branch divergence need diagnosis or recovery.
reviewing-code-quality
Reviews a diff or module for slipping standards, favoring deletion over rearranging, and ends in one honest verdict. Use when a change risks oversized files, needless layers, feature logic leaking into shared code, or clever indirection. Do not use for a trivial obvious edit, or for a check that is only about whether…
review-deep
Drive the deep-review phase of an automated PR review. Consumes the walkthrough, runs the deterministic deep-review workflow (parallel lenses → adversarial validation → code-enforced threshold/caps), drafts the surviving findings, and completes the review run.
review-orchestrator
Drive an automated PR review. Produces a structured walkthrough plus inline draft comments via the octomux review CLI. NEVER posts to GitHub directly — publishing is human-gated.
audit
Use for PACEflow internal full audit: run five independent review agents, verify evidence from code/tests/logs, de-duplicate findings, and produce a severity-ranked report for release gates or comprehensive code review.
readonly-review
Run a no-network read-only review practice with two explorer steps and one synthesis step.