RNA4219

shipyard CLIでplanからpublishまでを実行し、manual acceptanceとpublish approvalで安全に停止・再開する。

shipyard-cpの実行可能CLIでtask作成、dispatch、状態確認を行う。

Use when operating shipyard-cp and deciding how to mix local Qwen 4B or 27B runtimes with external providers. Apply it when the goal is to save tokens, move prep work to local models, reserve premium providers for high-risk stages, or define model roles…

Experiment Gate リポジトリで Claude 系エージェントが作業するときの最小ガイド。GateRequest の確認、`python -m experiment_gate` CLI 実行、`config/defaults.yaml` や `--set` による閾値調整、LLM 接続確認、README と実装整合の点検が必要なときに使う。

Experiment Gate リポジトリで Codex が作業するときの最小ガイド。Gate CLI の改善、`run_gate()` 周辺の修正、スコアリングや Persona 評価の確認、`python -m experiment_gate` の実行検証、設定反映や受け入れ確認を行うときに使う。

手動ブラックボックス前提で、仕様・受入条件・変更点・不具合履歴・自動テスト証跡から、根拠付きテスト観点、リスク優先度、手動テストケース、工数見積り、品質ゲート判定、Go/No-Go brief を作る。Use when Codex needs to design or review manual black-box test plans, derive equivalence/boundary/decision/state coverage, triage missing oracles, assess…

Maintain and release the harness-auto-test-evidence HATE repository. Use when Codex is asked to update HATE release docs, README files, product readiness claims, platform CLI evidence, major OSS validation records, codemap freshness, release candidate packs,…

Maintain and verify the ..\agent-taskstate repository. Use when working in this repo to review requirements, fix implementation gaps, align code with docs, tests, and schema, update README guidance, or prepare acceptance, commit, and push. Triggers include…

code-to-gate品質解析CLIツールのSkill。リポジトリ品質分析、検出ルール、ポリシー評価、リリース判定の整合性を保つ場合に使用。

workflow-cookbookのLLM活動ログをagent-protocols Evidence契約へ接続するSkill。StructuredLogger、evidence_bridge、plugin_loader/config、sample、schema、testsの整合性を保つ場合に使用。

Use when working in the aituber-protocols repository on protocol requirements, schemas, examples, test vectors, reference implementation, acceptance checks, README updates, or cross-file consistency between specs and code.

Build a planning-ready roadmap request from workflow-approved problem material and run Roadmap Design Skill validation or planning. Use when a user has already decided the case should be roadmaped, wants to assemble `problem_statement` / `insights` /…

Use when working in the tracker-bridge-materials repository on typed_ref canonicalization, tracker integration, issue_cache/entity_link/sync_event behavior, context rebuild flows, docs consistency, acceptance checks, or test updates. Trigger when the user…