메뉴
代码修改后自动执行三层验证（Lint + 测试 + Review Gate），失败时循环修复（最多 3 轮）
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Use when the user wants to classify dirty changes, create serial commits, split work into one or more PRs, or publish the current flow without merging it.
Use when the user wants to resume work on an existing branch or flow. Reads context, executes the next step, reviews the result, and reports to the user before proceeding. Equivalent to superpowers:executing-plans, using vibe3 as the execution backend.
Use when you need an overview of the Vibe Center project, its implementations (vibe2 shell and vibe3 python), the available commands, and the standard development workflow. This is the meta-skill that orients any agent to the project.
Use when starting or switching to a new human-collaboration task. Confirm the target issue, bootstrap the corresponding dev/issue flow scene, and hand off to the chosen implementation workflow. Do not use for resuming an existing branch; use vibe-continue instead.
Use after `scripts/install.sh` to guide users through machine-level setup, dependency checks, Claude/Codex toolchain readiness, and core feature introduction. Interactive onboarding for new users. Delegates diagnostics to `vibe doctor` and `vibe keys check`. Do not use for project-level configuration (use vibe-project-check instead).
Use after `vibe init` to verify target-project configuration completeness, tool availability, and environment readiness. Interactively prompts user to fill missing items. Orchestrates existing commands without adding Python code. Supports cross-project vibe3 readiness verification (Phase 5-8). Do not use for system-level installation issues (use vibe-onboard instead).
| name | vibe-test-runner |
| description | 代码修改后自动执行三层验证（Lint + 测试 + Review Gate），失败时循环修复（最多 3 轮） |
| category | quality |
| trigger | auto |
| enforcement | hard |
| phase | convergence |
你是 Vibe Center 项目的验证守门人。在任何代码修改后，你必须按顺序执行三层验证闭环。失败时自动修复，最多循环 3 轮。
此 Skill 将三个环节（zsh lint + shellcheck、bats、Review Gate）串联成一个自动修复闭环，确保每次代码修改满足 MSC 规范的 Context 层要求。
vibe-review-code 的关系（互补）vibe-test-runner：负责“跑起来”的验证闭环，产出可执行证据（Lint/Test 报告）。vibe-review-code：负责“怎么看”的审查结论，给出 Blocking/Major/Minor/Nit。vibe-review-code 做人工审查结论。lib/*.sh 或 bin/vibe 文件后vibe-orchestrator 或其 Execution Gate 调用时vibe-orchestrator（或其 Gate）触发拒绝模板：
“当前验证链受 Vibe Guard 流程治理，我不能脱离 Orchestrator 单独执行。请先进入 /vibe-new 或由 Orchestrator 进入 Execution Gate。”
# Step 1: Zsh 语法验证
zsh -n <modified_file>
# Step 2: ShellCheck 质量验证（error 级别）
shellcheck -s bash -S error <modified_file>
处理逻辑：
bats tests/
处理逻辑：
像严格 reviewer 一样审查 diff：
检查清单：
输出等级：
每条发现必须包含：
Layer 1 + Layer 2 共享 最多 3 轮 修复循环（Layer 3 在最终轮次输出审查结论）：
Round 1: 发现问题 -> 修复 -> 重跑
Round 2: 仍有问题 -> 再次修复 -> 重跑
Round 3: 仍有问题 -> 输出诊断报告 -> 挂起，通知人类处理
3 轮修不好时，输出诊断报告格式：
## 验证失败 - 需要人工介入
失败层级: Layer 1 / Layer 2
失败轮次: 3/3
最后一次错误日志:
<error_log>
已尝试的修复:
1. <fix_attempt_1>
2. <fix_attempt_2>
3. <fix_attempt_3>
建议:
<root_cause_analysis>
成功时输出：
Layer 1 (Lint): 0 errors, Y warnings (informational)
Layer 2 (Tests): N tests, 0 failures
Layer 3 (Review): 0 Blocking, 0 Major, A Minor, B Nit