スキル executor-architecture

📦

executor-architecture

Name: executor-architecture
Author: atri10

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

Document Architecture Before Writing Specifications

Vague architecture leaves teams with incompatible interfaces and undocumented decisions. This skill structures architecture, decisions, contracts, and component designs for human review before specification.

対応: Claude Codex Code(CC)

⚠️ 50 不十分

Maintain this Skill? Claim it ⚠️ 50 不十分

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-architecture" from https://skillstore.io/skills/atri10-executor-architecture.md and its manifest at https://skillstore.io/api/skills/atri10-executor-architecture/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-architecture.md 署名済み manifest GET /api/skills/atri10-executor-architecture/manifest 署名済み lockfile GET /api/skills/atri10-executor-architecture/lockfile

テストする

「executor-architecture」を使用しています。 Document architecture for tenant placement using the accepted per-cell storage approach.

期待される結果:

Components: placement policy, storage port, storage adapter, and HTTP adapter.
Boundary: policy depends only on policy-owned interfaces; the storage adapter implements the storage port.
Decision record: compare per-cell storage with shared storage and state the accepted operational costs.
Review: confirm failure responses and interface contracts before recording architecture approval.

「executor-architecture」を使用しています。 Two tasks disagree about the placement interface. Define the contract before implementation.

期待される結果:

Operations: list candidate cells, bind a tenant, and release a tenant.
Types: define identifier formats, capacity units, and timestamp precision.
Errors: specify unavailable storage, existing placement, and a cell that stopped accepting requests.
Lifecycle: require initialization, document concurrent calls, and name both consuming tasks.

「executor-architecture」を使用しています。 A storage decision no longer satisfies measured requirements. Replace it without losing history.

期待される結果:

New decision record: describe the changed constraint and the evidence that triggered reconsideration.
Alternatives: compare viable replacements and state migration costs.
History: link the successor and predecessor while preserving the original decision body.
Dependencies: update affected architecture references and initiative index entries.

セキュリティ監査

中リスク

v1 • 2026/10/5 バージョン別レポートを開く

Most alerts are Markdown formatting, fixed sibling paths, or document identifiers rather than executable backticks, path traversal exploits, or reconnaissance. Required helper executions remain confirmed with low confidence because their implementations are outside the five reviewed files. Verification can record a passed phase before human approval; no evidence found of exfiltration, credential theft, or audit-directed prompt injection.

スキャンされたファイル

1,191

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 (1)

中

SKILL.md:156-175 SKILL.md:409-423

Verification Can Record Approval Before Human Review

Verification says "Run these in this session and cite their output at the gate" and includes a command recording architecture as passed. This conflicts with "On approval" and the instruction to stop at the gate. An agent could record unapproved architecture as accepted, weakening downstream approval controls.

The pre-gate verification list explicitly includes a passed-state mutation, while the procedure requires prior human approval. Actual enforcement cannot be checked without the helper implementation.

機能レビュー項目 (14)

これらは、このスキルに期待される可能性のある実際のローカル機能であるため、レビューが必要ですが、確認済みの悪意ある動作としてはカウントされません。

中

SKILL.md:49-52

Ruby/shell backtick execution

```bash

The fenced block instructs executing a sibling helper to resolve an initiative and mutate its phase log. Its implementation is unavailable, so execution safety remains uncertain.

中

SKILL.md:78-80

Ruby/shell backtick execution

```bash

The block requires executing the sibling architecture ID allocator. Its source is outside this package, so its filesystem effects and execution safety cannot be verified.

中

SKILL.md:100-102

Ruby/shell backtick execution

```bash

The block requires executing the sibling ADR ID allocator. Its source is unavailable, so its filesystem effects and execution safety cannot be verified.

中

SKILL.md:119-121

Ruby/shell backtick execution

```bash

The block requires executing the sibling interface ID allocator. Its implementation is outside this package, so execution safety cannot be verified.

中

SKILL.md:168-170

Ruby/shell backtick execution

```bash

The block invokes a sibling helper to persist a passed architecture phase after approval. The helper implementation is unavailable, leaving execution effects uncertain.

中

SKILL.md:179-182

Ruby/shell backtick execution

```bash

The block requires sibling helpers for design phase entry and ID allocation. Their implementations are unavailable, so execution safety cannot be verified.

中

SKILL.md:189-191

Ruby/shell backtick execution

```bash

The block invokes a sibling helper to persist design approval. Its implementation is unavailable, so its execution effects cannot be independently verified.

中

SKILL.md:195-197

Ruby/shell backtick execution

```bash

The block invokes a sibling helper to persist a skipped design phase. Its implementation is unavailable, leaving filesystem and execution effects uncertain.

中

SKILL.md:288

Ruby/shell backtick execution

| `frozen` | A plan is executing against it | Do **not** edit. Write a new IFCE that supersedes it,

The frozen-interface rule instructs executing a sibling ruling helper with plan and decision arguments. Its implementation and argument handling are unavailable, so execution safety remains uncertain.

中

SKILL.md:343-347

Ruby/shell backtick execution

`../executor/scripts/exec-initiative phase INIT-0004 design skipped "<reason>"`

This instruction invokes a sibling phase-log helper with a supplied reason. The implementation and argument handling are unavailable, leaving execution safety uncertain.

中

SKILL.md:357-358

Ruby/shell backtick execution

1. Allocate a new ID: `../executor/scripts/exec-id INIT-0004 ADR`.

The supersession procedure requires executing the sibling ADR allocator. Its implementation is outside the reviewed package, so execution safety cannot be verified.

中

SKILL.md:413-414

Ruby/shell backtick execution

1. `../executor/scripts/exec-store-check` — no finding for this

Verification requires executing the sibling store-check helper. Its source is unavailable, so claimed validation behavior and execution safety cannot be verified.

中

SKILL.md:420-422

Ruby/shell backtick execution

4. `../executor/scripts/exec-scan-secrets docs/executor/<INIT>-<slug>/architecture`

Verification requires running a sibling helper over architecture documents to scan secrets. Its unavailable implementation could access document contents, so safety cannot be established.

中

SKILL.md:422-427

Ruby/shell backtick execution

5. `../executor/scripts/exec-initiative phase <INIT> architecture passed "..."`

Verification invokes a sibling helper that records a passed architecture phase. Its implementation is unavailable; the separate semantic finding addresses premature approval recording.

リスク要因

📁 ファイルシステムへのアクセス (24)

⚙️ 外部コマンド (50)

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

HTMLバッジ

<a href="https://skillstore.io/skills/atri10-executor-architecture?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/atri10-executor-architecture/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-architecture.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-architecture security audit report (audit version 1) [Author version unspecified]. Skillstore. https://skillstore.io/skills/atri10-executor-architecture/audits/1

BibTeX形式の引用

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

Define System Boundaries

Turn an accepted approach into component responsibilities, dependency rules, failure responses, and reviewable architecture.

Coordinate Shared Interfaces

Define exact operations, types, errors, and ordering so independently assigned tasks can integrate against the same contract.

Preserve Decision History

Record alternatives and reversal criteria, then supersede outdated decisions without deleting their original reasoning.

これらのプロンプトを試す

Draft an Architecture Overview

For [initiative], use the charter and accepted approach to draft the architecture overview. State constraints, component responsibilities, boundaries, and open questions. Stop for human review before recording approval.

Record an Architecture Decision

For [initiative], document the decision about [topic]. Compare viable alternatives, accepted costs, reversibility, and observable reversal criteria. Link the decision to the architecture and prepare it for review.

Define a Shared Interface

For [initiative], define the interface between [provider] and [consumer]. Specify exact operations, types, errors, ordering, concurrency, lifecycle, stability, and consumers. Identify missing information without inventing implementation details.

Supersede a Frozen Contract

For [initiative], evaluate replacing [frozen interface] because [new evidence]. Draft its successor, affected consumer updates, decision rationale, and component design changes. Preserve prior document bodies and request approval before recording phase completion.

ベストプラクティス

Provide the initiative charter, accepted approach, and measured constraints before drafting architecture.
Review helper implementations and dependency paths before execution; keep validation read-only until explicit human approval.
Keep contracts exact, index documents in the same change, and preserve superseded reasoning.

回避

Starting specifications or recording passed phases before human approval.
Letting separate tasks invent shared signatures, error contracts, or ordering rules.
Editing historical decision bodies or frozen interfaces instead of writing linked successor documents.

よくある質問

What does this skill produce?

It guides architecture documents, decision records, interface contracts, and component designs. Outputs include linked metadata, review summaries, and explicit skip records.

Does it write implementation code?

No. It permits signatures and type declarations as interface contract text, but implementation belongs to later phases.

Can it run independently?

The full workflow requires the sibling executor skill, shared references, helper scripts, and an existing initiative inside a Git repository.

Which AI tools are supported?

The marketplace metadata lists Claude, Codex, and Claude Code. The package contains Markdown instructions and templates rather than tool-specific executable integrations.

When can documents be skipped?

Skip decisions without viable alternatives, designs with obvious internals, or interfaces without shared boundaries. Record reasons explicitly; skipping architecture requires human waiver.

What security precautions are needed?

Review required helper scripts before running them. Keep credentials out of documents, and record passed phases only after explicit human approval.