fix: manageCloudRun 零配置部署能力边界不清晰，错误返回缺少 actionable 信息，导致 Agent 无法正确完成部署

[binggg]

Attribution issue

• issueId: issue_moj17gxm_j5qv9w
• category: tool
• canonicalTitle: manageCloudRun 零配置部署能力边界不清晰，错误返回缺少 actionable 信息，导致 Agent 无法正确完成部署
• representativeRun: atomic-js-app-zero-config-deploy/2026-04-28T20-08-14-02c40a

Automation summary

• root_cause: The cloudrun-development SKILL.md explicitly said "Do NOT use for Frontend-only projects with no backend service", which blocked agents from using CloudRun to deploy frontend projects. The skill also had no zero-config deployment guidance — the deploy example always included a full serverConfig, leading agents to believe manual configuration was required. The error messages from manageCloudRun deploy only handled the "already deploying" case, providing generic advice for all other failures, giving agents no actionable next steps when deployment failed.
• changes: (1) SKILL.md — Removed "Frontend-only projects" from the "Do NOT use for" list, added a note that CloudRun supports frontend deployment with auto-detection. Added a full "Zero-config deployment" section explaining when/how to omit serverConfig, auto-detection rules for serverType, and a step-by-step frontend deployment flow. Added zero-config deploy example as the primary/recommended example, moved the full serverConfig example to secondary. Added "Verify deployment after deploy" examples with queryCloudRun(action="detail") and getDeployLog. Expanded Troubleshooting with actionable causes an

Changed files

• config/.claude/skills/cloudrun-development/SKILL.md
• config/source/skills/cloudrun-development/SKILL.md
• doc/prompts/auth-web.mdx
• doc/prompts/cloudrun-development.mdx
• mcp/src/tools/cloudrun.ts

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.

[binggg]

Attribution post-PR evaluation

• visibility: internal identifiers, run ids, and private links are intentionally omitted
• attempt: 1
• eval_scope: primary_only
• overall: FAILED
• summary: at least one planned evaluation case failed
• updated_at: 2026-04-28T20:56:17.223Z

Cases

• [FAILED] — primary — evaluation failed