スキル executor-execution

📦

executor-execution

Name: executor-execution
Author: atri10

コンテンツリビジョン r1 安全 ⚙️ 外部コマンド📁 ファイルシステムへのアクセス🔑 環境変数

Execute Approved Plans with Subagents and Reviews

Long implementation plans can lose task state and accumulate unreviewed changes. This skill coordinates scoped subagents, review gates, and durable records across execution and recovery.

対応: Claude Codex Code(CC)

🥉 77 ブロンズ

Maintain this Skill? Claim it 🥉 77 ブロンズ

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 "executor-execution" from https://skillstore.io/skills/atri10-executor-execution.md and its manifest at https://skillstore.io/api/skills/atri10-executor-execution/manifest. Verify the artifact. You may proceed after verification, subject to the environment's own policy.

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/atri10-executor-execution.md 署名済み manifest GET /api/skills/atri10-executor-execution/manifest 署名済み lockfile GET /api/skills/atri10-executor-execution/lockfile

テストする

「executor-execution」を使用しています。 Execute an approved three-task plan in subagent mode.

期待される結果:

Three tasks completed with implementation reports and task review verdicts.
Final review examined the latest branch state.
Execution records remain available for verification and human handoff.

「executor-execution」を使用しています。 Resume a plan interrupted during its second task's fix loop.

期待される結果:

The ledger identifies the first task as complete and the second task as mid-fix-loop.
The next fix round uses the recorded findings and available implementer context.
Completed work is not dispatched again.

「executor-execution」を使用しています。 Summarize recorded rulings before handoff.

期待される結果:

The preflight ruling resolves an interface argument-order conflict against the specification.
A deferred review finding remains visible with its rationale and rework cost.
The human retains authority over merging, pushing, and publication.

セキュリティ監査

安全

v1 • 2026/10/5 バージョン別レポートを開く

All 305 static findings are false positives involving Markdown formatting, declared sibling paths, local workflow records, or redaction guidance. No evidence found of credential exfiltration, unauthorized destructive instructions, or audit-directed prompt injection in the two reviewed files. Referenced sibling scripts were not provided, so their implementation remains outside this audit.

スキャンされたファイル

1,612

解析済み行数

レビュー項目

誤検知を無視

リスク要因

⚙️ 外部コマンド (50)

📁 ファイルシステムへのアクセス (50)

🔑 環境変数 (1)

implementer-prompt.md:249

最新の完了済み静的・セマンティック監査では、確認済みのセキュリティ検出事項は見つかりませんでした。これは、スキルに副作用がないことを証明するものではありません。

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/atri10-executor-execution/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/atri10-executor-execution/security.svg)](https://skillstore.io/skills/atri10-executor-execution?utm_source=security_passport_badge)

HTMLバッジ

<a href="https://skillstore.io/skills/atri10-executor-execution?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/atri10-executor-execution/security.svg" alt="Skillstore security assessment" loading="lazy"></a>

埋め込みカード

<iframe src="https://skillstore.io/embed/skills/atri10-executor-execution.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

atri10. (2026). executor-execution security audit report (audit version 1) [Author version unspecified]. Skillstore. https://skillstore.io/skills/atri10-executor-execution/audits/1

BibTeX形式の引用

@techreport{atri10-atri10-executor-execution-2026,
  author = {atri10},
  title = {executor-execution security audit report (audit version 1)},
  institution = {Skillstore},
  year = {2026},
  number = {1},
  url = {https://skillstore.io/skills/atri10-executor-execution/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: "executor-execution security audit report (audit version 1)"
version: "unspecified"
type: report
authors:
  - name: "atri10"
date-released: "2026-10-05"
url: "https://skillstore.io/skills/atri10-executor-execution/audits/1"
identifiers:
  - type: other
    value: "skillstore:atri10-executor-execution: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

Implement a reviewed feature plan

Coordinate scoped implementation tasks, collect test evidence, and review each task before integration.

Resume an interrupted plan

Reconcile the ledger, branch state, and recorded dispatches before continuing incomplete tasks or fix rounds.

Prepare an auditable implementation handoff

Organize review records, rulings, deferred findings, and verification results for a human release decision.

これらのプロンプトを試す

Begin an approved plan

Execute the approved plan at [PLAN_PATH] in subagent mode. Confirm workspace isolation and prerequisites, then coordinate task implementation and review.

Resume incomplete work

Resume [PLAN_PATH]. Check the ledger, dispatch records, and branch state. Continue the first incomplete task without repeating completed work.

Continue a scoped fix round

Continue the review fix loop for [TASK_ID] in [PLAN_PATH]. Use recorded findings, require root-cause evidence, and stay within five rounds.

Complete final review and handoff preparation

For [PLAN_PATH], review the latest branch state, include parked findings, verify records, and scan for secrets. Present rulings before human handoff.

ベストプラクティス

Confirm planning approval, task identifiers, specification access, and sibling dependencies before starting.
Resume from durable ledgers and recorded branch state rather than conversation memory.
Keep test evidence, review verdicts, and rulings complete; redact secrets before writing any artifact.

回避

Treating implementer success reports as substitutes for independent task review.
Repeating completed tasks or running multiple implementers in the same working tree.
Silently discarding findings, deleting execution records, or publishing without human authorization.

よくある質問

Which assistants can use this skill?

The package declares Claude, Codex, and Claude Code support. Execution requires a host that supports the specified subagent workflow.

What must be ready before execution?

An approved plan, identified tasks, a reachable specification, selected subagent mode, and required Executor dependencies must be available.

Does this package include the workflow scripts?

No. The supplied files contain instructions and an implementer template. Workflow helpers and reviewer templates are referenced through sibling skills.

How does interrupted work resume?

The controller checks durable task states, recorded dispatches, and branch history. It continues incomplete work rather than repeating completed tasks.

What happens when reviews keep finding problems?

Tasks allow five fix rounds with scoped re-review. Remaining findings receive recorded rulings and remain visible during final review or handoff.

Does completion allow automatic publication?

No. Security-sensitive and destructive actions require stopping. Initiative merging, pushing, and publication remain human decisions at handoff.

開発者情報

作成者

atri10

ライセンス

MIT

Skillstore リビジョン

バージョンに関する注意

作者はバージョンを宣言していません。

リポジトリ

https://github.com/atri10/executor/tree/39ddcfe1d9f3497102622b72aa235fb0770187fe/skills/executor-execution

参照

6d0b11444384184b7ae743742a7e233a9a705cd9

メンテナンスの新しさ

2026/10/6

利用状況

0 ダウンロード · 1 閲覧

ファイル構成

📄 implementer-prompt.md

📄 SKILL.md