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 rjmurillo/ai-agents --skill reviewer-findingsgit clone --depth 1 https://github.com/rjmurillo/ai-agentsWrote 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/rjmurillo/ai-agents/reviewer-findings)<a href="https://agentmods.dev/skills/rjmurillo/ai-agents/reviewer-findings"><img src="https://agentmods.dev/badge/skills/rjmurillo/ai-agents/reviewer-findings/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/rjmurillo/ai-agents/reviewer-findings"><img src="https://agentmods.dev/badge/skills/rjmurillo/ai-agents/reviewer-findings.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to high
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- high Prompt Injection · line 66 Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.Fix: Audit all comments and invisible characters. Remove any instructions that direct the agent to perform unauthorized actions. Use plain, reviewable content.
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 | $0.00122 | $0.05065 |
| Opus 5.5 | $0.00049 | $0.02026 |
| Sonnet 5.5 | $0.00024 | $0.01013 |
| Haiku 4.5 | $0.00012 | $0.00507 |
Grade A, and why
reviewer-findings 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 22d 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 — 344 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Reviewer Findings
A review finding is evidence that someone looked, not proof that they were right. Verify it before you act on it. This skill is the discipline for consuming a finding; it does not produce one.
Triggers
address this review commentthe bot flagged thishandle this findinga sub-agent reported this
Also fires without any of those phrases when a bot or human leaves a review comment, when an adversarial sub-agent or a model on another family returns a report, or when you inherit findings from a compaction note, a handoff, or a prior session.
The three claims
A finding is up to three separate claims stacked together. Each carries its own evidence, and each can fail while the others hold.
| Claim | Example | What it is worth on arrival |
|---|---|---|
| Verdict | "this fails on Windows" | A hypothesis until you reproduce it |
| Diagnosis | "because that call is POSIX-only" | An account of the repro, not proof of it |
| Prescription | "guard the call or skip on Windows" | A claim with its own decay |
A correct verdict can carry a wrong diagnosis. A correct diagnosis can carry an incomplete fix. Accepting the bundle because the top line is right is how a wrong fix lands with a green review.
The dangerous shape: reviewing the motivation, not the result
A finding can be internally consistent with a pre-fix version of the tree: the reviewer reconstructs the problem a commit solved and reports that, so every prescription matches what the commit already did. This kind of finding is maximally plausible, because it is a true statement about the world one commit ago, and it defeats the obvious check: verifying the recommendation ("is this the right value?") confirms the finding. Only verifying the premise ("does the file actually say this, right now?") exposes it.
Treat premise verification as its own triage outcome, not a variant of an ordinary defect:
| Premise | What settles it | Disposition |
|---|---|---|
| True | The check matching the claim's shape (single-line current state: git grep -n -F; multi-line current state: whole-block comparison; provenance or history: git log -S; see Process below) matches the claim |
Confirmed; proceed to fix |
| False | That same shape-matched check refutes the claim, or the prescribed fix is already present and the verdict does not reproduce on its own | Declined (refuted); reply with the file, line, commit, and evidence, no code change |
| Unverifiable | None of the above settles it | Unreproduced; reply with what you tried, leave the thread open |
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.
- 22d ago Changed · +6 lines 5f9167e8084a
- 1mo ago First seen · 338 lines · 122 tokens per session scan A 04e77c280908
reviewer-findings is a skill published in the GitHub repository rjmurillo/ai-agents (48 stars, last pushed today), licensed MIT. It adds 122 tokens to every session and 5,065 once invoked, about $0.0005 per session on Opus 5.5. 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 2026-09-03.
Other skills, from other repositories
install-gemini-code-review-action
Installs, configures, or upgrades Dazbo's PR code review & issue triage GitHub Action (gemini-review-action) in a target repository. Use when the user wants to set up, configure, or update automated AI code reviews, issue triaging, or Gemini model workflows for their GitHub repository.
review-deep
Review change vs spec, plan, code quality heuristics, and repo rules; append remediation phase if needed.
review-remote
Review open PR/MR and publish findings.
apply-review-feedback
Classify and apply accepted PR/MR review feedback.
azure-devops-automation
CI/CD pipelines, infrastructure as code, and deployment automation for Azure workloads.
architecture-audit
Comprehensive project consistency review across code, documentation, diagrams, and configuration.