📦
git-workflow
78一貫した Git ワークフローを実行する
作成者 Doyajin174
Git 作業は、不明確なコミットや検証漏れに陥りがちです。このスキルは、Conventional Commits、ブランチ命名、チェック、プッシュ準備をガイドします。
DevOps＆インフラ 表示
スキル agy-worker
📦
agy-worker
v0.21.0 コンテンツリビジョン r8 高リスク ⚙️ 外部コマンド🌐 ネットワークアクセス📁 ファイルシステムへのアクセス⚡ スクリプトを含む🔑 環境変数
Delegate Repository Work Safely
Repository delegation can consume time and make verification inconsistent. This skill sends bounded work to agy, then keeps review, checks, repair, and acceptance with Codex.
対応: Codex
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 "agy-worker" from https://skillstore.io/skills/cagdasyurekli-agy-worker.md and its manifest at https://skillstore.io/api/skills/cagdasyurekli-agy-worker/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。
テストする
「agy-worker」を使用しています。 Explore the authentication flow and identify the files that handle token validation.
期待される結果:
- Mapped the authentication entry points and token validation path.
- Listed evidence files and marked unverified assumptions for Codex review.
「agy-worker」を使用しています。 Add parser error-path tests under tests/parser and verify the change.
期待される結果:
- Implemented the focused tests within the approved path.
- Ran the relevant test command and reported the final candidate status.
「agy-worker」を使用しています。 Audit the repository workflow and repair bounded failures.
期待される結果:
Returned a structured project result with the candidate diff, verification evidence, remaining limits, and delivery recommendation.
セキュリティ監査
高リスクv8 • 2026/9/20 バージョン別レポートを開く
Most static matches are benign implementation, validation, documentation, schema, or control-character checks. Confirmed risks include provider and verifier execution, environment inheritance, keychain handling, settings access, and non-sandboxed provider authority. Static review was capped at 400/991 representative findings; omitted static matches are unconfirmed, so automatic publishing stays disabled until manual review.
92
スキャンされたファイル
39,230
解析済み行数
50
レビュー項目
0
誤検知を無視
確認済みのセキュリティ上の懸念 (7)
高
Certificate/key files
keychain_path=prepared.keychain.path if prepared.keychain is not None else None,
The code binds a prepared keychain path or keychain preferences file, which is a sensitive-resource capability. It is used for native isolation, but compromise of the boundary could expose credential material.
高
Certificate/key files
if prepared.keychain is None or prepared.keychain_preferences is None:
The code binds a prepared keychain path or keychain preferences file, which is a sensitive-resource capability. It is used for native isolation, but compromise of the boundary could expose credential material.
高
Certificate/key files
if _bind_keychain(prepared.keychain.path) != prepared.keychain:
The code binds a prepared keychain path or keychain preferences file, which is a sensitive-resource capability. It is used for native isolation, but compromise of the boundary could expose credential material.
高
Certificate/key files
prepared.keychain_preferences.path, modes={0o600}, limit=MAX_PROFILE_BYTES,
The code binds a prepared keychain path or keychain preferences file, which is a sensitive-resource capability. It is used for native isolation, but compromise of the boundary could expose credential material.
高
Certificate/key files
) != prepared.keychain_preferences:
The code binds a prepared keychain path or keychain preferences file, which is a sensitive-resource capability. It is used for native isolation, but compromise of the boundary could expose credential material.
高
Certificate/key files
prepared.keychain is not None or prepared.keychain_preferences is not None
The code binds a prepared keychain path or keychain preferences file, which is a sensitive-resource capability. It is used for native isolation, but compromise of the boundary could expose credential material.
高
External Provider Transmission Without Host Isolation
The skill can transmit a whole disposable worktree to Google or Gemini, while default session mode retains normal user filesystem and network authority without host containment.
The documented workflow explicitly states both the external transmission boundary and the lack of sandbox or native containment in default session mode.
機能レビュー項目 (50)
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。
高
Hidden file in home directory
p = os.path.expanduser("~/.gemini/antigravity-cli/settings.json")
The account-phase helper expands and reads the user settings file under ~/.gemini. This accesses provider configuration that may contain sensitive permissions or account settings.
高
Process spawn
return spawn(job, "initial", resume=False, foreground=True)
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Process spawn
return spawn(job, "initial", resume=False, foreground=False)
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Process spawn
return spawn(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Process spawn
return spawn(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python os.exec variants
os.execvpe(command[0], command, dict(os.environ))
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.Popen
process = subprocess.Popen(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.Popen
process = subprocess.Popen(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.Popen
child = subprocess.Popen(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.Popen
process = subprocess.Popen(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.Popen
process = subprocess.Popen(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.Popen
controller_process = subprocess.Popen(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.Popen
process = subprocess.Popen(argv, cwd=cwd, env=git_env(), stdin=subprocess.DEVNULL, stdout=subprocess
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.Popen
process = subprocess.Popen(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.Popen
process = subprocess.Popen(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.run
return subprocess.run(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.run
subprocess.run(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.run
subprocess.run(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.run
provider_checked = subprocess.run(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.run
canonical_checked = subprocess.run(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.run
subprocess.run(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.run
result = subprocess.run(command, cwd=cwd, env=git_env(), stdin=subprocess.DEVNULL, stdout=subprocess
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.run
completed = subprocess.run(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.run
completed = subprocess.run(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.run
proc = subprocess.run(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.run
proc = subprocess.run(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.run
proc = subprocess.run(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.run
proc = subprocess.run(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.run
return subprocess.run(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.run
proc = subprocess.run(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.run
proc = subprocess.run(verify_cmd, check=False)
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
高
Python subprocess.run
fin_proc = subprocess.run(
The runtime launches provider, Git, or verification processes. This is required by the skill, but it creates a real command-execution boundary that must remain approval-bound.
中
Hidden file access
p = os.path.expanduser("~/.gemini/antigravity-cli/settings.json")
The account-phase helper expands and reads the user settings file under ~/.gemini. This accesses provider configuration that may contain sensitive permissions or account settings.
中
Unix shell invocation
/bin/bash -c "${verify_specs[$i]}"
The verifier invokes a shell or defines a shell supervisor. Shell execution can run arbitrary commands, so this is a real execution boundary even with the documented controls.
中
Unix shell invocation
/bin/bash -c "${verify_specs[$i]}"
The verifier invokes a shell or defines a shell supervisor. Shell execution can run arbitrary commands, so this is a real execution boundary even with the documented controls.
中
Unix shell invocation
SCRIPT_SHELLS = frozenset({"/bin/bash", "/bin/sh"})
The verifier invokes a shell or defines a shell supervisor. Shell execution can run arbitrary commands, so this is a real execution boundary even with the documented controls.
中
Unix shell invocation
"/bin/sh", "-c", supervisor, "bounded-git-supervisor",
The verifier invokes a shell or defines a shell supervisor. Shell execution can run arbitrary commands, so this is a real execution boundary even with the documented controls.
中
Unix shell invocation
"/bin/bash",
The verifier invokes a shell or defines a shell supervisor. Shell execution can run arbitrary commands, so this is a real execution boundary even with the documented controls.
低
Python environment access
environment = os.environ.copy()
The code reads or copies the process environment for child execution. Environment variables can contain credentials, so filtering must remain complete and approval-bound.
低
Python environment access
if name in os.environ:
The code reads or copies the process environment for child execution. Environment variables can contain credentials, so filtering must remain complete and approval-bound.
低
Python environment access
(os.fsencode(name), b"\0", os.fsencode(os.environ[name]), b"\0")
The code reads or copies the process environment for child execution. Environment variables can contain credentials, so filtering must remain complete and approval-bound.
低
Python environment access
os.execvpe(command[0], command, dict(os.environ))
The code reads or copies the process environment for child execution. Environment variables can contain credentials, so filtering must remain complete and approval-bound.
低
Python environment access
environment = os.environ.copy()
The code reads or copies the process environment for child execution. Environment variables can contain credentials, so filtering must remain complete and approval-bound.
低
Python environment access
environment = {name: os.environ[name] for name in allowed if name in os.environ}
The code reads or copies the process environment for child execution. Environment variables can contain credentials, so filtering must remain complete and approval-bound.
低
Python environment access
configured = os.environ.get("XDG_STATE_HOME")
The code reads or copies the process environment for child execution. Environment variables can contain credentials, so filtering must remain complete and approval-bound.
低
Python environment access
home_text = os.environ.get("HOME")
The code reads or copies the process environment for child execution. Environment variables can contain credentials, so filtering must remain complete and approval-bound.
低
Python environment access
environment = dict(os.environ)
The code reads or copies the process environment for child execution. Environment variables can contain credentials, so filtering must remain complete and approval-bound.
低
Python environment access
env = dict(os.environ)
The code reads or copies the process environment for child execution. Environment variables can contain credentials, so filtering must remain complete and approval-bound.
低
Python environment access
os.environ.get("AGY_WORKER_LOG_DIR") or (state_path.parent / "logs")
The code reads or copies the process environment for child execution. Environment variables can contain credentials, so filtering must remain complete and approval-bound.
低
Python environment access
os.environ.get("AGY_WORKER_LOG_DIR") or (SCRIPTS.parent / "logs")
The code reads or copies the process environment for child execution. Environment variables can contain credentials, so filtering must remain complete and approval-bound.
リスク要因
⚙️ 外部コマンド (50)
README.md:56 references/PROJECT_LIFECYCLE_AND_VERIFICATION.md:71 references/PROJECT_LIFECYCLE_AND_VERIFICATION.md:80 references/PROJECT_LIFECYCLE_AND_VERIFICATION.md:81 references/PROJECT_LIFECYCLE_AND_VERIFICATION.md:82 references/PROJECT_LIFECYCLE_AND_VERIFICATION.md:252 references/PROJECT_LIFECYCLE_AND_VERIFICATION.md:253 references/PROJECT_LIFECYCLE_AND_VERIFICATION.md:287 references/PROJECT_LIFECYCLE_AND_VERIFICATION.md:288-290 runtime/agy-worker.sh:14 runtime/agy-worker.sh:15 runtime/agy-worker.sh:1262 runtime/agy-worker.sh:1263 runtime/agy-worker.sh:1300 runtime/agy-worker.sh:25 runtime/agy-worker.sh:30 runtime/agy-worker.sh:237 runtime/agy-worker.sh:464 runtime/agy-worker.sh:473-475 runtime/agy-worker.sh:497 runtime/agy-worker.sh:498 runtime/agy-worker.sh:499 runtime/agy-worker.sh:603 runtime/agy-worker.sh:604 runtime/agy-worker.sh:649 runtime/agy-worker.sh:650 runtime/agy-worker.sh:651 runtime/agy-worker.sh:652 runtime/agy-worker.sh:663 runtime/agy-worker.sh:685 runtime/agy-worker.sh:690 runtime/agy-worker.sh:692 runtime/agy-worker.sh:696 runtime/agy-worker.sh:720-725 runtime/agy-worker.sh:901-906 runtime/agy-worker.sh:980 runtime/agy-worker.sh:1009-1011 runtime/agy-worker.sh:1043-1054 runtime/agy-worker.sh:1225 runtime/agy-worker.sh:1226 runtime/agy-worker.sh:1306 runtime/agy-worker.sh:15-1262 runtime/benchmark.sh:5 runtime/codex-usage-report.sh:5 runtime/delegation-policy.sh:5 runtime/doctor.sh:21 runtime/doctor.sh:32 runtime/doctor.sh:37-38 runtime/doctor.sh:85-86 runtime/doctor.sh:146-147
🌐 ネットワークアクセス (43)
references/PROJECT_LIFECYCLE_AND_VERIFICATION.md:25 runtime/compat/agy-version-manifest.json:20 runtime/compat/agy-version-manifest.json:64 runtime/compat/agy-version-manifest.json:111 runtime/compat/agy-version-manifest.json:163 runtime/compat/agy-version-manifest.json:212 runtime/compat/agy-version-manifest.json:257 runtime/compat/agy-version-manifest.json:277 runtime/compat/agy-version-manifest.json:303 runtime/compat/agy-version-manifest.json:347 runtime/compat/model-effort-matrix.schema.json:2 runtime/compat/model-effort-matrix.schema.json:3 runtime/compat/version-manifest.schema.json:2 runtime/schemas/benchmark-plan.schema.json:1 runtime/schemas/benchmark-result.schema.json:1 runtime/schemas/delegation-policy.schema.json:2 runtime/schemas/evidence-receipt.schema.json:2 runtime/schemas/job-state.schema.json:2 runtime/schemas/model-evidence-campaign-advisory-preview.schema.json:2 runtime/schemas/model-evidence-campaign-advisory-summary.schema.json:2 runtime/schemas/model-evidence-campaign-aggregate-preview.schema.json:2 runtime/schemas/model-evidence-campaign-aggregate.schema.json:2 runtime/schemas/model-evidence-campaign-evaluation.schema.json:2 runtime/schemas/model-evidence-campaign-plan.schema.json:2 runtime/schemas/model-evidence-campaign-record.schema.json:2 runtime/schemas/model-evidence-campaign-record.schema.json:156 runtime/schemas/model-intelligence-advisory.schema.json:2 runtime/schemas/model-intelligence-evidence.schema.json:2 runtime/schemas/model-intelligence-evidence.schema.json:85 runtime/schemas/model-recommendation.schema.json:2 runtime/schemas/model-selection.schema.json:2 runtime/schemas/swebench-workflow-study-advisory.schema.json:1 runtime/schemas/swebench-workflow-study-report.schema.json:1 runtime/schemas/worker-result.provider.schema.json:2 runtime/schemas/worker-result.schema.json:2 runtime/schemas/workflow-state.schema.json:2 runtime/scripts/codex_usage_report.py:726 runtime/scripts/compatibility.py:245 runtime/scripts/compatibility.py:246 runtime/scripts/feedback-triage.py:31 runtime/scripts/model_evidence_campaign.py:28 runtime/scripts/model_intelligence.py:26 SKILL.md:36
📁 ファイルシステムへのアクセス (50)
references/PROJECT_LIFECYCLE_AND_VERIFICATION.md:81 references/PROJECT_LIFECYCLE_AND_VERIFICATION.md:82 references/PROJECT_LIFECYCLE_AND_VERIFICATION.md:252 references/SECURITY_AND_COMPATIBILITY.md:213 references/SECURITY_AND_COMPATIBILITY.md:42 runtime/agy-worker.sh:1262 runtime/agy-worker.sh:1262 runtime/agy-worker.sh:237 runtime/agy-worker.sh:663 runtime/agy-worker.sh:673 runtime/agy-worker.sh:685 runtime/agy-worker.sh:897 runtime/agy-worker.sh:1011 runtime/agy-worker.sh:1030 runtime/agy-worker.sh:1200 runtime/agy-worker.sh:1201 runtime/agy-worker.sh:939 runtime/doctor.sh:21 runtime/doctor.sh:24 runtime/doctor.sh:32 runtime/doctor.sh:38 runtime/doctor.sh:86 runtime/doctor.sh:147 runtime/doctor.sh:156 runtime/doctor.sh:157 runtime/doctor.sh:172 runtime/doctor.sh:198 runtime/doctor.sh:199 runtime/doctor.sh:225 runtime/doctor.sh:230 runtime/doctor.sh:231 runtime/doctor.sh:237 runtime/doctor.sh:252 runtime/doctor.sh:253 runtime/doctor.sh:301 runtime/doctor.sh:302 runtime/doctor.sh:304 runtime/doctor.sh:315 runtime/doctor.sh:322 runtime/doctor.sh:468 runtime/doctor.sh:470 runtime/doctor.sh:478 runtime/doctor.sh:479 runtime/doctor.sh:493 runtime/doctor.sh:506 runtime/doctor.sh:527 runtime/doctor.sh:543 runtime/doctor.sh:563 runtime/doctor.sh:579 runtime/doctor.sh:224
⚡ スクリプトを含む (22)
runtime/agy-worker.sh:1435 runtime/agy-worker.sh:1436 runtime/agy-worker.sh:1448 runtime/agy-worker.sh:1451 runtime/agy-worker.sh:1462 runtime/agy-worker.sh:1493 runtime/scripts/agy_dispatch_containment.py:809-810 runtime/scripts/agy_dispatch_worktree.py:971 runtime/scripts/agy_dispatch_worktree.py:4548 runtime/scripts/agy_dispatch.py:2998 runtime/scripts/agy_dispatch.py:46 runtime/scripts/agy_dispatch.py:3292 runtime/scripts/benchmark.py:88-96 runtime/scripts/benchmark.py:97-100 runtime/scripts/evidence_receipt.py:29-37 runtime/scripts/evidence_receipt.py:38-41 runtime/scripts/evidence_report.py:25-34 runtime/scripts/evidence_report.py:35-38 runtime/scripts/job_lifecycle.py:35-38 runtime/scripts/job_lifecycle.py:39-44 runtime/scripts/job_lifecycle.py:45-54 runtime/scripts/model-recommendation.py:12-18
🔑 環境変数 (15)
runtime/qa-gate.sh:120 runtime/scripts/agy_dispatch_verification.py:151 runtime/scripts/agy_dispatch_worktree.py:1676 runtime/scripts/evidence_receipt.py:1115 runtime/scripts/evidence_receipt.py:1117 runtime/scripts/evidence_receipt.py:1521 runtime/scripts/evidence_receipt.py:955 runtime/scripts/feedback-triage.py:360 runtime/scripts/model_selection.py:98 runtime/scripts/workflow.py:764 runtime/scripts/workflow.py:771 runtime/scripts/workflow.py:803 runtime/scripts/workflow.py:1168 runtime/scripts/workflow.py:1242 runtime/scripts/workflow.py:1667
監査者: codex 監査履歴を表示 →
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。
レポートリンクをコピー
https://skillstore.io/skills/cagdasyurekli-agy-worker/audits/8?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_reportMarkdownバッジ
[](https://skillstore.io/skills/cagdasyurekli-agy-worker?utm_source=security_passport_badge)HTMLバッジ
<a href="https://skillstore.io/skills/cagdasyurekli-agy-worker?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/cagdasyurekli-agy-worker/security.svg" alt="Skillstore security assessment" loading="lazy"></a>埋め込みカード
<iframe src="https://skillstore.io/embed/skills/cagdasyurekli-agy-worker.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
cagdasyurekli. (2026). agy-worker security audit report (audit version 8) [Author version 0.21.0]. Skillstore. https://skillstore.io/skills/cagdasyurekli-agy-worker/audits/8BibTeX形式の引用
@techreport{cagdasyurekli-cagdasyurekli-agy-worker-2026,
author = {cagdasyurekli},
title = {agy-worker security audit report (audit version 8)},
institution = {Skillstore},
year = {2026},
number = {8},
url = {https://skillstore.io/skills/cagdasyurekli-agy-worker/audits/8},
note = {Author version 0.21.0}
}CITATION.cff
cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "agy-worker security audit report (audit version 8)"
version: "0.21.0"
type: report
authors:
- name: "cagdasyurekli"
date-released: "2026-09-20"
url: "https://skillstore.io/skills/cagdasyurekli-agy-worker/audits/8"
identifiers:
- type: other
value: "skillstore:cagdasyurekli-agy-worker:audit:8"
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 82
アーキテクチャ
100
保守性
87
コンテンツ
67
コミュニティ
91
仕様準拠
作成できるもの
Explore an unfamiliar repository
Ask agy to map code paths or investigate a focused question, then review the evidence with Codex.
Implement a bounded change
Delegate a feature, refactor, or test change within an approved scope and verify the resulting diff independently.
Run a broad project workflow
Use project mode for larger audits or repairs while preserving review, checks, and delivery decisions with Codex.
これらのプロンプトを試す
Start a repository exploration
Use agy-worker to inspect the repository and explain the code paths related to [topic]. Read only the approved scope and report evidence.
Implement a focused task
Use agy-worker to implement [change] under [paths]. Preserve existing behavior, run relevant checks, and summarize the final diff.
Verify a delegated change
Use agy-worker to review the candidate for [task]. Run independent checks, identify gaps, and repair only within the approved scope.
Run a controlled project workflow
Use agy-worker project mode for [objective]. Require approved provider scope, explicit model and budget, independent verification, and an evidence-based delivery decision.
ベストプラクティス
- Approve the exact provider-readable content, model, budget, and isolation mode before launch.
- Prefer provider scopes and argv verification, then inspect the final diff independently.
- Keep secrets, raw logs, controller state, and unrelated private files outside approved content.
回避
- Do not treat installation as permission to transmit repository content.
- Do not trust worker envelopes, reported tests, or provider claims without independent checks.
- Do not use whole-worktree or shell verification modes without explicit scope and risk approval.
よくある質問
What does agy-worker delegate?
It delegates repository exploration, implementation, testing, and bounded repair to Google Antigravity CLI.
What remains with Codex?
Codex retains scope decisions, diff review, independent verification, repair decisions, and final acceptance.
Does the skill transmit repository files?
It can transmit approved scoped files or an explicitly approved whole worktree to the external provider.
Is the provider session a sandbox?
No. Default session mode retains normal user filesystem and network authority without native host containment.
Which hosts are supported?
The package supports OpenAI Codex CLI hosts with Bash, Python 3, Git, and agy available.
How should I verify results?
Inspect the actual diff and run driver-owned checks in an isolated verification copy before delivery.
開発者情報
作成者
cagdasyurekliライセンス
MIT
作者バージョン
v0.21.0
Skillstore リビジョン
r8
参照
73b58aaf0aa4d29844f1b4c2acf4f530fc1ae7b9
メンテナンスの新しさ
2026/9/20
利用状況
6 ダウンロード · 19 閲覧
ファイル構成
📁 agents/
📄 PROJECT_LIFECYCLE_AND_VERIFICATION.md
📄 SECURITY_AND_COMPATIBILITY.md
📁 runtime/
📁 agents/
📁 v1/
📁 tasks/
📁 compat/
📄 agy-model-effort-matrix.json
📄 agy-model-effort-matrix.sha256
📄 agy-models-inventory-binding.json
📄 agy-models-inventory-binding.sha256
📄 model-effort-matrix.schema.json
📄 version-manifest.schema.json
📄 job.sh
📁 schemas/
📄 benchmark-result.schema.json
📄 delegation-policy.schema.json
📄 evidence-receipt.schema.json
📄 model-evidence-campaign-advisory-preview.schema.json
📄 model-evidence-campaign-advisory-summary.schema.json
📄 model-evidence-campaign-aggregate-preview.schema.json
📄 model-evidence-campaign-aggregate.schema.json
📄 model-evidence-campaign-evaluation.schema.json
📄 model-evidence-campaign-plan.schema.json
📄 model-evidence-campaign-record.schema.json
📄 model-intelligence-advisory.schema.json
📄 model-intelligence-evidence.schema.json
📄 model-recommendation.schema.json
📄 swebench-workflow-study-advisory.schema.json
📄 swebench-workflow-study-plan.schema.json
📄 swebench-workflow-study-report.schema.json
📄 worker-result.provider.schema.json
📁 scripts/
📄 agy_dispatch_verification.py
📁 scripts/
📄 SKILL.md
関連スキル
すべて表示📦
ariadne-loop
81検証可能なエージェントループ仕様を作成
作成者 zhangzeyu99-web
曖昧なコーディングエージェントのタスクは、しばしば方向がずれたり、作業を繰り返したり、証拠なしに停止したりします。このスキルは、リクエストを検証器、停止ルール、ハンドオフパケットを備えた境界付きのループ仕様に変換します。
コーディング＆開発 表示
📦
verification-protocol
80証拠に基づいてエージェントの完了を検証
作成者 CleanExpo
AIエージェントは証拠なしに作業を完了とマークできるため、誤った完了リスクが生じます。このスキルは、タスク完了が受け入れられる前に独立した証拠を要求します。
生産性 表示
📦
plan-writing
79焦点を絞った実装計画を作成する
作成者 sickn33
複数ステップのコーディング作業は、曖昧で大規模になり、検証しにくくなることがあります。このスキルは、Claude、Codex、Claude Code が、明確なタスク、順序、チェックを備えた短い計画を作成するのに役立ちます。
コーディング＆開発 表示
📦
executing-plans
79チェックポイント付きで実装計画を実行
作成者 obra
作業が別の AI セッションに移ると、実装計画は焦点を失うことがあります。このスキルは、Claude、Codex、Claude Code に対して、バッチ実行、検証、レビューのチェックポイントを通じて作業を進めるよう導きます。
コーディング＆開発 表示
📦
verification-before-completion
79報告前に完了の主張を検証する
作成者 CodingCossack
エージェントが記憶や自信に頼ると、完了の主張は誤ることがあります。このスキルでは、成功を報告する前に新しい証拠を求めます。
コーディング＆開発 表示