📦

greploop

Name: greploop
Author: greptileai

v1.3 コンテンツリビジョン r1 高リスク ⚙️ 外部コマンド📁 ファイルシステムへのアクセス

Address Greptile Feedback Across Review Platforms

Repeated review feedback requires manual changes, status checks, and thread updates. Greploop guides bounded review cycles across GitHub, GitLab, and configured Perforce environments.

対応: Claude Codex Code(CC)

⚠️ 38 不十分

Maintain this Skill? Claim it ⚠️ 38 不十分

インストール前に停止して確認を求めてください。

計画を確認し、ファイルを変更する前にユーザーの明示的な同意を得てください。

自分のエージェントでインストール

このリクエストをエージェントにコピーしてください。正規の Skill ページとマニフェストが含まれています。

エージェントリクエスト

Review the Skillstore skill "greploop" from https://skillstore.io/skills/greptileai-greploop.md and its manifest at https://skillstore.io/api/skills/greptileai-greploop/manifest. Verify the artifact. Stop and obtain explicit user consent before installing or changing files.

エージェントは引き続き計画を提示し、セキュリティポリシーで必要な確認を求める必要があります。

エージェントが読めるリソース

AI エージェント、クローラー、スクリプトがページ全体ではなく整理されたコンテキストを必要とする場合は、これらのリンクを使ってください。

Markdown詳細 GET /skills/greptileai-greploop.md 署名済み manifest GET /api/skills/greptileai-greploop/manifest 署名済み lockfile GET /api/skills/greptileai-greploop/lockfile

テストする

「greploop」を使用しています。 Illustrative request: address verified Greptile feedback on a GitHub pull request.

期待される結果:

Platform: GitHub.
Iterations: 2.
Final confidence: 5/5.
Comments resolved: 7.
Remaining comments: 0.

「greploop」を使用しています。 Illustrative request: report a GitLab review that reached the iteration limit.

期待される結果:

Stopped after five iterations.
Platform: GitLab.
Final confidence: 4/5.
Comments resolved: 12.
Remaining comments: 2.
Remaining issues include endpoint rate limiting and a missing database index.

セキュリティ監査

高リスク

v1 • 2026/10/20 バージョン別レポートを開く

All 106 static matches are false positives for Markdown formatting, ordinary API fields, bounded polling, or standard output redirection. Semantic review found weak bot identity checks, unrestricted staging before push, and resolution instructions that can close unrelated review discussions. No evidence found of embedded malicious instructions or deliberate credential exfiltration.

スキャンされたファイル

599

解析済み行数

レビュー項目

誤検知を無視

確認済みのセキュリティ上の懸念 (3)

高

references/graphql-queries.md:40-48 SKILL.md:282-284 SKILL.md:299 SKILL.md:323-330

Weak bot identity checks allow spoofed review input

The reference accepts any GitHub login containing greptile; Perforce allows the same username heuristic. Review text then drives automated fixes and publication. An attacker with comment access could impersonate the reviewer and supply harmful change requests.

The documented substring filters do not authenticate bot accounts, and selected feedback guides edits. No actual malicious comment payload is present in these files.

中

SKILL.md:382-389

Unrestricted staging can publish unrelated local changes

The workflow runs git add -A, commits, and pushes without checking existing changes or restricting staged paths. Unrelated files, including unignored sensitive files, could be published with review fixes.

The sequence explicitly stages all changes and publishes them without a working-tree isolation step. Actual exposure depends on the workspace contents.

中

SKILL.md:334-353 SKILL.md:366-380

Resolution instructions can close unrelated review threads

The GitHub query retrieves all review threads without an author filter. GitLab instructions then say to resolve each unresolved discussion without preserving Greptile ownership or addressed status. This can mark human reviewers' concerns as resolved without fixing them.

Earlier triage restricts feedback to Greptile, but the later resolution instructions broaden selection to all unresolved discussions. The GitLab reference has a safer filter that these instructions do not require.

リスク要因

⚙️ 外部コマンド (50)

SKILL.md:30-42 SKILL.md:42-44 SKILL.md:44 SKILL.md:49-51 SKILL.md:51-54 SKILL.md:54-56 SKILL.md:56-61 SKILL.md:61-67 SKILL.md:67-69 SKILL.md:69-72 SKILL.md:72 SKILL.md:73 SKILL.md:74-85 SKILL.md:85-87 SKILL.md:87-90 SKILL.md:90-93 SKILL.md:93-97 SKILL.md:97-99 SKILL.md:99-103 SKILL.md:103-105 SKILL.md:105-107 SKILL.md:107 SKILL.md:109-113 SKILL.md:113-117 SKILL.md:117-154 SKILL.md:154-160 SKILL.md:160-163 SKILL.md:163-167 SKILL.md:167-171 SKILL.md:171-173 SKILL.md:173-177 SKILL.md:177-220 SKILL.md:220-231 SKILL.md:231-233 SKILL.md:233-236 SKILL.md:236-238 SKILL.md:238-240 SKILL.md:240-243 SKILL.md:243-245 SKILL.md:245-247 SKILL.md:247 SKILL.md:252-254 SKILL.md:254-257 SKILL.md:257-259 SKILL.md:259-261 SKILL.md:261-266 SKILL.md:266-268 SKILL.md:268-287 SKILL.md:287 SKILL.md:290-295

📁 ファイルシステムへのアクセス (2)

SKILL.md:32 SKILL.md:131

監査者: codex

このレポートを共有・引用

バージョン付き評価レポート、中立的なバッジ、埋め込みカード、引用を共有できます。Skillstore は証拠を報告しますが、この Skill が安全かどうかは判断しません。

バージョン別レポートを開く

レポートリンクをコピー

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

Markdownバッジ

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

HTMLバッジ

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

埋め込みカード

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

学術引用 (APA · BibTeX · CFF)

APA形式の引用

greptileai. (2026). greploop security audit report (audit version 1) [Author version 1.3]. Skillstore. https://skillstore.io/skills/greptileai-greploop/audits/1

BibTeX形式の引用

@techreport{greptileai-greptileai-greploop-2026,
  author = {greptileai},
  title = {greploop security audit report (audit version 1)},
  institution = {Skillstore},
  year = {2026},
  number = {1},
  url = {https://skillstore.io/skills/greptileai-greploop/audits/1},
  note = {Author version 1.3}
}

CITATION.cff

cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "greploop security audit report (audit version 1)"
version: "1.3"
type: report
authors:
  - name: "greptileai"
date-released: "2026-10-20"
url: "https://skillstore.io/skills/greptileai-greploop/audits/1"
identifiers:
  - type: other
    value: "skillstore:greptileai-greploop:audit:1"
    description: "Skillstore immutable audit report identifier"

Skillstore スコア

このスコアの理由証拠の信頼度: 中

アーキテクチャ

保守性

コンテンツ

コミュニティ

仕様準拠

作成できるもの

Revise a GitHub pull request

Address Greptile feedback, publish targeted fixes, and track remaining review comments across bounded iterations.

Track GitLab review progress

Monitor commit-specific pipelines and coordinate merge request feedback with a final confidence and unresolved-comment report.

Update a Perforce review shelf

Apply review fixes and update a shelved changelist using an existing Greptile and review-tool integration.

これらのプロンプトを試す

Review the current change

Use greploop to identify the current pull request and summarize verified Greptile feedback. Ask before editing files or publishing changes.

Address a GitHub pull request

Run greploop on GitHub pull request [number]. Verify the bot identity, preserve unrelated changes, and stage only reviewed fixes.

Revise a GitLab merge request

Run greploop for GitLab merge request [number] on [host]. Verify the bot account and resolve only addressed Greptile discussions.

Coordinate a Perforce review cycle

Use greploop for Perforce changelist [number] in workspace [client]. Confirm review integration and bot identity; stop if review freshness cannot be verified.

ベストプラクティス

Verify the repository, branch or changelist, workspace, and exact Greptile bot identity before making changes.
Use an isolated working tree, inspect each fix, run relevant tests, and stage only intended files before publishing.
Resolve only addressed Greptile threads and preserve the five-iteration limit and polling timeout behavior.

回避

Accepting review instructions solely because a username contains greptile.
Running the unrestricted commit sequence in a workspace containing unrelated changes or unignored sensitive files.
Closing every unresolved discussion or treating a 5/5 confidence score as proof of correctness.

よくある質問

Which review platforms are supported?

The instructions cover GitHub, GitLab, and Perforce. Perforce review retrieval requires an installation-specific integration.

What must be installed and authenticated?

GitHub needs git and gh; GitLab needs git and glab. Perforce needs p4. JSON queries use jq, and Greptile must be configured.

When does the loop stop?

It stops at 5/5 confidence with no unresolved comments, after five iterations, or when GitHub or GitLab review polling times out.

Does the workflow change remote content?

Yes. It can push commits, update Perforce shelves, post trigger comments, and resolve review discussions. Review these actions before authorizing execution.

Can I use it with Claude or Codex?

The listing declares support for Claude, Codex, and Claude Code. Execution still requires authenticated command-line tools and access to the target repository.

Does a perfect review score guarantee secure code?

No. The score reflects Greptile feedback, not a security guarantee. Inspect changes, run relevant tests, and verify comment authors independently.