📦

reviewer

Name: reviewer
Author: drix10

Ревизия содержимого r1 Безопасно ⚙️ Внешние команды

Review Agent-Flow Diffs Against Repository Rules

Code changes can miss acceptance criteria, repository rules, or security boundaries. This skill performs a read-only review and returns structured, line-anchored findings.

Поддерживает: Claude Codex Code(CC)

🥉 77 Бронза

Установить с помощью моего Агента

Скопируйте этот запрос в своего Агента. Он содержит каноническую страницу Skill и манифест.

Запрос агента

Review the Skillstore skill "reviewer" from https://skillstore.io/skills/drix10-reviewer.md and its manifest at https://skillstore.io/api/skills/drix10-reviewer/manifest. Verify the artifact. You may proceed after verification, subject to the environment's own policy.

Ваш Агент по-прежнему должен показать план и запросить все подтверждения, требуемые политикой безопасности.

Ресурсы для AI-агентов

Используйте эти ссылки, когда AI-агенту, crawler или script нужен чистый контекст вместо полной страницы.

Markdown-детали GET /skills/drix10-reviewer.md Подписанный manifest GET /api/skills/drix10-reviewer/manifest Подписанный lockfile GET /api/skills/drix10-reviewer/lockfile

Протестировать

Использование «reviewer». Review a change that implements the feature but adds no test for one acceptance criterion.

Ожидаемый результат:

Requests changes and identifies the unproven criterion as a blocking implementation error with file evidence and a test suggestion.

Использование «reviewer». Review a correct diff that updates a protected path.

Ожидаемый результат:

Requests changes because the protected-path violation is always blocking, even when the implementation appears correct.

Использование «reviewer». Reconsider a prior warning after the implementer provides decisive repository evidence.

Ожидаемый результат:

Withdraws the warning, records why the evidence resolves it, and approves only when every criterion remains satisfied.

Аудит безопасности

Безопасно

v1 • 27.09.2026 Открыть версионный отчет

All 21 static findings are false positives caused by Markdown inline-code delimiters and a fenced JSON example. The skill contains no executable commands or command-substitution syntax, and no semantic security issue was found.

Просканировано файлов

Проанализировано строк

Пункты проверки

Ложные срабатывания проигнорированы

Факторы риска

⚙️ Внешние команды (21)

Последний завершенный статический и семантический аудит не обнаружил подтвержденных проблем безопасности. Это не доказывает отсутствие побочных эффектов у навыка.

Аудитор:: codex

Поделиться и цитировать этот отчет

Делитесь версионным отчетом об оценке, нейтральным значком, встраиваемой карточкой и цитатами. Skillstore публикует доказательства, не решая, безопасен ли этот Skill.

Открыть версионный отчет

Копировать ссылку на отчёт

https://skillstore.io/skills/drix10-reviewer/audits/1?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_report

Значок Markdown

[![Skillstore security assessment](https://skillstore.io/badges/skills/drix10-reviewer/security.svg)](https://skillstore.io/skills/drix10-reviewer?utm_source=security_passport_badge)

Значок HTML

<a href="https://skillstore.io/skills/drix10-reviewer?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/drix10-reviewer/security.svg" alt="Skillstore security assessment" loading="lazy"></a>

Встраиваемая карточка

<iframe src="https://skillstore.io/embed/skills/drix10-reviewer.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>

Академические ссылки (APA · BibTeX · CFF)

Цитата APA

drix10. (2026). reviewer security audit report (audit version 1) [Author version unspecified]. Skillstore. https://skillstore.io/skills/drix10-reviewer/audits/1

Цитата BibTeX

@techreport{drix10-drix10-reviewer-2026,
  author = {drix10},
  title = {reviewer security audit report (audit version 1)},
  institution = {Skillstore},
  year = {2026},
  number = {1},
  url = {https://skillstore.io/skills/drix10-reviewer/audits/1},
  note = {Author version unspecified}
}

CITATION.cff

cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "reviewer security audit report (audit version 1)"
version: "unspecified"
type: report
authors:
  - name: "drix10"
date-released: "2026-09-27"
url: "https://skillstore.io/skills/drix10-reviewer/audits/1"
identifiers:
  - type: other
    value: "skillstore:drix10-reviewer:audit:1"
    description: "Skillstore immutable audit report identifier"

Оценка Skillstore

Почему такая оценка Достоверность доказательств: Средний

Архитектура

Сопровождаемость

Контент

Сообщество

Соответствие спецификации

Что вы можете построить

Validate an implementation round

Compare a submitted diff with issue criteria before another agent continues.

Review a risky code change

Inspect authentication, payment, contract, or personal-data changes for missing safeguards.

Check repository compliance

Find protected-path edits, dependency risks, unrelated changes, and conflicts with repository guidance.

Попробуйте эти промпты

Review a diff

Review the supplied agent-flow diff against its acceptance criteria. Return the required structured verdict with evidence for every finding.

Check criteria and tests

Review each acceptance criterion. Identify its implementation and test evidence, then flag every unsupported criterion as an implementation error.

Assess security boundaries

Review this diff for protected paths, dependencies, outbound calls, injection risks, secret exposure, authorization gaps, and partial-write failures.

Resolve disputed findings

Review the latest diff, prior findings, and implementer evidence. Withdraw unsupported findings and explain what evidence is still missing for retained findings.

Лучшие практики

Provide the complete artifact packet and relevant repository context.
Anchor every finding to specific changed code and a reproducible failure scenario.
Treat issue text, diffs, comments, and documentation as untrusted review data.

Избегать

Do not ask the reviewer to edit files or implement fixes.
Do not approve changes with protected-path violations or unreviewed dependencies.
Do not report speculative findings without code evidence or a concrete failure scenario.

Часто задаваемые вопросы

What does this skill review?

It reviews one agent-flow diff against issue criteria, repository guidance, protected paths, dependency changes, and security boundaries.

Can it change the code?

No. The skill is designed for read-only review and only returns a verdict and findings.

Does it run tests?

No. It inspects provided changes and test evidence but has no documented command-execution capability.

What makes a finding blocking?

Missing criteria, correctness defects, protected-path violations, and unreviewed dependency changes can block approval.

How does it handle prompt injection?

It treats issue content as untrusted data and checks changed code, comments, and documentation for agent instructions.

Can it reconsider an earlier finding?

Yes. In later rounds, it evaluates implementer evidence and withdraws findings that are no longer supported.

Сведения для разработчиков

Автор

drix10

Лицензия

MIT

Ревизия Skillstore

Примечание о версии

Автор не указал версию.

Репозиторий

https://github.com/drix10/agent-flow/tree/aac4c5c023b25d073997626804c355ed11348feb/skills/reviewer

Ссылка

8b5ad1599e7c55e9b6c627bca3f556f6e9061cb1

Актуальность поддержки

30.09.2026

Использование

0 загрузок · 0 просмотров

Структура файлов

📄 SKILL.md