[cli-tools-test] CLI exploratory testing: windows strict dry-run gate fails on deprecated missing-tool key

オープン 初心者向け
  • 何か問題がありますか？
#75,317 コメント 0 件 リアクション 0 件 担当者 0 名 GitHub で見る

メンテナーはふだん 1 日以内に返信

まだ誰も着手していません。

評価

難易度
1/5
見積もり時間
1〜3時間
初心者へのやさしさ
68/100
issue の種類
バグ
明瞭さ
明確に書かれている
活発さ
活発
技術スタック
github-actions, go
領域
ci-cd, tooling

調査の方向性

windowsのワークフロー設定から始め、廃止予定のセーフ出力キーmissing-toolを特定します。issueでは、置き換え先としてreport-missing-toolが示されています。通常のメンテナンス手順でキーを更新し、windowsの厳密なドライランコンパイルを再実行して、ワークフローの結果とバッチゲートの両方が通ることを確認します。

索引モデルが issue の本文から書いたものです。

説明

automation cli-tools testing
Overview

Bounded CLI exploration on 2026-11-06 found a repository workflow configuration problem: windows emits a lock but fails its strict dry-run gate because it uses the deprecated missing-tool safe-output key. Severity: Low; observed once. This is not evidence of a compiler crash.

Reproduction and diagnostics
  1. Invoke agenticworkflows compile with {"workflows":["windows"],"dry_run":true,"strict":true}.
  2. Inspect both the workflow result and batch gate.

Expected: the trusted sample passes strict compilation, or clearly reports required configuration maintenance. Actual: workflow valid: true, followed by batch valid: false and gate: failed. Consumers must inspect the batch result.

Redacted diagnostics

The missing-tool safe-output key is deprecated. Use report-missing-tool instead. The diagnostic appears twice within compiler_warning; no line number was supplied. Shellcheck passed with zero added findings. The emitted lock metadata includes strict: true, dry_run: true, and frontmatter/body hashes.

Passed checks
  • Status discovery returned workflow names, engines, triggers, compilation state, and dependencies (~4.6s); exhaustive enumeration comparison was not performed.
  • Three logs queries, each with count: 3: last day (3 runs), weekly-editors-health-check over seven days (1 run), and copilot over the last day (3 runs). Exposed JSON summaries were readable; sampled names, engines, timestamps, and count limits matched filters. Count-limited responses clearly reported partial coverage. Times: ~9.8s, 6.0s, and 12.5s; the engine query exceeded the 10s aspirational target in this single observation.
  • Nonexistent workflow returned an empty result and actionable guidance (~0.1s).
  • Successful-run audit matched logs metadata and included job/step timing, resource metrics, MCP/tool usage, safe-output summary, and network counts (~5.7s). Parent success did not establish error-free tool execution: grader results reported tool failures.
  • Failed-run audit returned failure metadata matching logs and an actionable incomplete-task finding (~13.3s). Full report review was limited by output truncation.
  • Exactly three strict dry-run compile attempts: weekly-editors-health-check passed (~2.1s); windows failed its gate (~2.4s); archie was rejected (~1.2s). Two emitted locks were inspected for metadata, jobs, triggers, and safe-output jobs. No lock was executed or published.
Blocked or unfinished required checks

archie was rejected at .github/workflows/archie.md:1:1 because dry-run cannot safely process its pre-agent steps and custom MCP servers. No retry, fixture edit, or bypass was attempted. Thus three passing compile gates and full lock verification were not achieved. Audit report completeness beyond the visible bounded output remains unverified. A separate incomplete signal records these limits.

Not tested (expected exclusion)
  • Synthetic malformed YAML, invalid tool configurations, and missing imports: trusted-target restriction.
  • Private raw logs and independent artifact completeness: outside session scope.
  • Newly compiled lock execution: explicitly prohibited.
  • Additional historical/in-progress audits: bounded two-run sample.
  • Redundant date variants, future/old-date queries, invalid-ID audit, incremental compile repetition, and resource profiling: optional variants omitted to keep the sample bounded.
Next actions

Migrate the windows workflow key to report-missing-tool through the normal maintenance path and revalidate. Preserve the archie safety rejection. Existing-run evidence for weekly-editors-health-check comes from SHA cf88bbcb56e5844b95101d742daf79067dfb1300, not the newly emitted dry-run lock. No repository-wide health claim is made.

References:

Generated by 🧪 Daily Cli Tools Tester · codex · gpt60 · ~103 AIC · ⌖ ~62 AIC · ⊞ 21.3K · ◷

  • expires on Nov 12, 2026, 9:23 PM UTC-08:00
主要言語
Go
スター
5.6k
フォーク
615
平均マージ
5時間 21分
マージ済み PR（30日）
792

環境構築

Codespaces で開く

このプロジェクトの開発コンテナを、あなたの GitHub アカウントでブラウザ上に起動します。

はじめの一歩

  1. issue を最後まで読み、次にプロジェクトのコントリビューションガイドを読みます。
  2. 着手することを issue にコメントします — 二人が同じ作業をするのを防げます。
  3. リポジトリをフォークし、ブランチを切って変更します。
  4. issue 番号を参照したプルリクエストを送ります。

github/gh-aw のほかの issue

github/gh-aw の issue をすべて見る

似ている issue

Go の issue をもっと見る

新しい issue をメールで受け取る

初心者向けの GitHub issue を短くまとめたダイジェスト。