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 patforna/auto-task --skill review-codegit clone --depth 1 https://github.com/patforna/auto-taskWrote 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/patforna/auto-task/review-code)<a href="https://agentmods.dev/skills/patforna/auto-task/review-code"><img src="https://agentmods.dev/badge/skills/patforna/auto-task/review-code/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/patforna/auto-task/review-code"><img src="https://agentmods.dev/badge/skills/patforna/auto-task/review-code.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 | $0.00051 | $0.02416 |
| Opus 5.5 | $0.00020 | $0.00966 |
| Sonnet 5.5 | $0.00010 | $0.00483 |
| Haiku 4.5 | $0.00005 | $0.00242 |
Grade C, and why
review-code scanned grade C with 1 finding 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 1mo 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.
Hidden instructionshighPrompt injection
Directives inside HTML comments, invisible characters or bidirectional overrides are read by the model and not by the person reviewing the file.
<!-- Distilled from a review of the code-review research literature (May 2026). This file is the distilled instruction, not the bibliography. --> How it starts
The opening of the file, as written. The whole thing — 152 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Code Review
Usage
/at:review-code [diff target] [further instructions]
Diff target: a commit, a commit range, a base branch, PR ref, etc.
Goal
Surface the small set of findings a competent author would fix if they saw them, and capture mechanically-certain trivia as ready-to-apply patches.
Notes
- Most findings are expected to be around design and evolvability, not defects.
- Defects are the high-severity minority that gates the ship.
- You classify, you do not fix: emit the fix, never apply it.
Authorship Guard
This must be run by a different model, or at least in a fresh session, as same-context self-review is fundamentally flawed.
If you authored the diff under review in the current session, or there is no diff target:
- warn the user that same-context self-review is structurally unreliable and advise re-invoking from a fresh session.
- if declined, downgrade review scope to: convention, CLAUDE.md, project-rule conformance, the do-not-flag filter, and the autofix lane only. Do NOT run an open correctness / logic / security pass in this context. A user override may only further restrict scope — never re-authorise the open pass in the same context.
The Do-Not-Flag List
Drop, do not surface:
- Pre-existing issues — anything this diff did not introduce or worsen.
- Tool-owned — lint, formatting, type errors, test failures. The deterministic gate runs separately; never duplicate it.
- Correct-but-unusual — unconventional but provably correct is not a finding.
- Framework-handled — errors or validation the framework already guarantees.
- Speculative — anything you cannot pin to a
file:linewith a concrete failure mechanism. - Judgement-bound trivia — taste-level naming or structure where reasonable authors differ and nothing is measurably wrong.
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.
- 1mo ago Changed · +1 lines 9fbfea9fea9e
- 1mo ago First seen · 151 lines · 51 tokens per session scan C cd593444e944
review-code is a skill published in the GitHub repository patforna/auto-task (11 stars, last pushed 1mo ago), licensed MIT. It adds 51 tokens to every session and 2,416 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it C with 1 finding (hidden instructions). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
code-review-excellence
Master effective code review practices to provide constructive feedback, catch bugs early, and foster knowledge sharing while maintaining team morale. Use when reviewing pull requests, establishing review standards, or mentoring developers.
multi-reviewer-patterns
Coordinate parallel code reviews across multiple quality dimensions with finding deduplication, severity calibration, and consolidated reporting. Use this skill when organizing multi-reviewer code reviews, calibrating finding severity, or consolidating review results.
babysit
Watch an open PR — fix failing CI, handle the straightforward review comments, and drive it to a mergeable state. Claude Code analog of Cursor's built-in /babysit. Use after opening a PR when the user wants the agent to shepherd it without re-prompting.
make-pr-easy-to-review
Prepare PRs for review by cleaning noisy history, improving PR descriptions, and adding reviewer guidance without changing code behavior. Use for "make this easy to review", "tidy this PR", "clean up commits", or "annotate the diff".
thermo-nuclear-code-quality-review
Run an extremely strict maintainability review for abstraction quality, giant files, and spaghetti-condition growth. Use for a thermo-nuclear code quality review, thermonuclear review, deep code quality audit, or especially harsh maintainability review.
interrogate
Use for "interrogate", "adversarial review", "multi-model review", "challenge this", "stress test this code", "find blind spots", or "tear this apart". Multiple LLM reviewers challenge changes from independent angles.