ci-cd-watchdog

CI/CD流水线智能运维助手，自动解析构建日志、定位根因、生成修复方案与回滚指令、输出事后复盘报告

Security Scan

The skill is an instruction-only CI/CD log analysis assistant whose requested inputs and behavior are consistent with its description and it does not request credentials, install software, or perform unexpected actions.

Purpose & Capability

Name/description (CI/CD log parsing, root-cause, repair/rollback suggestions, post-mortem) match the SKILL.md, examples.json and README; no unrelated binaries, env vars, or config paths are requested.

Instruction Scope

Runtime instructions operate only on the provided pipeline_log, optional ci_platform and repo_context and describe producing diagnostics, commands and a post-mortem. This is appropriate, but the skill purposefully emits operational commands and file paths — users must not run those blindly in production. The README does advise redacting secrets and using Dry-Run/approval flows.

Install Mechanism

No install spec and no code files — instruction-only skill. Nothing will be written to disk or downloaded during install.

Credentials

The skill requests no environment variables, credentials, or config paths. The optional repo_context may include repo/branch identifiers but does not require access tokens; this is proportionate to its stated function.

Persistence & Privilege

always is false and default autonomous invocation is allowed (platform default). The skill does not request persistent system-level presence or write other skills' configs.

Assessment

This skill is instruction-only and internally consistent with its CI/CD troubleshooting purpose, but exercise operational caution: 1) Do not paste unredacted secrets/tokens into pipeline_log — redact them as README recommends. 2) Treat generated repair and rollback commands as suggestions: review, run in staging, follow change-approval/Dry-Run procedures before applying in production. 3) Verify file paths, git commits, and environment assumptions in the output — an LLM can misdiagnose or hallucinate exact commands. 4) Because no code is installed, the skill itself does not exfiltrate data, but any sensitive information you include in inputs could be exposed in logs or downstream systems; avoid sharing secrets. If you need the skill to act directly against your CI/CD system (e.g., run git revert), prefer implementing a controlled adapter that requires explicit credentials and approval rather than pasting secrets into the skill inputs.

🐕 CI/CD 流水线智能运维助手

🎯 核心定位

将冗长构建日志转化为“阶段定位→根因诊断→修复命令→安全回滚→预防清单”的标准化运维闭环，缩短 MTTR（平均恢复时间）。

🔄 工作流指令

1. 日志解析：按时间轴切分阶段（Build/Test/Deploy/Notify），提取失败节点与关键错误堆栈。
2. 模式匹配：识别常见故障类型（依赖冲突/权限不足/磁盘已满/网络超时/配置漂移/代码语法）。
3. 方案生成：输出精准修复步骤（含文件路径/行号/环境变量），区分“临时绕过”与“根本解决”。
4. 回滚评估：若涉及生产/预发环境，生成安全回滚指令与影响面分析。
5. 复盘输出：生成 Post-Mortem 草稿与防复发 Checklist，按标准 Markdown 模板输出。

📤 输出模板

# 🔧 CI/CD 故障诊断报告

## 1. 失败定位
| 阶段 | 错误类型 | 关键日志行 | 触发条件 |
|:---|:---|:---|:---|
| ... | ... | `Line #...` | ... |

## 2. 根因分析与修复方案
- **根本原因**：...
- **修复步骤**：
  1. `...`
  2. `...`
- **验证命令**：`...`

## 3. 回滚与影响评估（如适用）
- **回滚指令**：`...`
- **影响范围**：...
- **数据风险提示**：...

## 4. 预防 Checklist
- [ ] 添加依赖锁定文件 (package-lock.json / poetry.lock)
- [ ] 增加集成测试覆盖关键路径
- [ ] 配置构建缓存策略与超时阈值
- [ ] 环境变量注入增加类型校验
> ⚠️ 所有命令已做安全过滤。生产环境操作请走变更审批流并执行 Dry-Run。