jkrandom-sudo/dsh-ci-doctor

CI 失败 —— 打开日志前即已诊断 —— DSH 插件:监视 GitHub Actions 新失败,将原始日志转化为结构化诊断卡,签名账本识别复发问题

Project Overview项目介绍

dsh-ci-doctor is a CI diagnostics plugin for DeepSeek Harness, exposing two agent tools against GitHub Actions. It parses raw job logs into normalized error signatures, failure categories, suspect files, and trimmed excerpts, and keeps a ledger of signatures so repeat failures are flagged on sight. The watch tool polls for newly failed runs in the background; the diagnose tool returns a markdown diagnosis card for a specific run. All access goes through the already-authenticated gh CLI and is strictly read-only — the plugin never pushes, merges, reruns, or writes to repositories. Requirement: an authenticated GitHub CLI.

dsh-ci-doctor 是 DeepSeek Harness 的 CI 诊断插件,通过两个 agent 工具接入 GitHub Actions。它将原始 job 日志解析为归一化的错误签名、失败分类、可疑文件与裁剪后的日志片段,并维护一张签名账本,重复失败会被即时标记。Watch 工具后台轮询新增失败运行,Diagnose 工具对单次运行输出 Markdown 诊断卡。仅通过已认证的 gh CLI 读取仓库状态,从不写入或触发任何操作。前提:已安装并登录 GitHub CLI。

Pre-install check安装前体检Compatibility · Security兼容性 · 安全性 1 warning1 项注意
  • Only 2 stars - very few users, little community feedback星标只有 2,几乎没人在用,遇到问题缺少社区反馈
DSH walks through these 9 checksDSH 会逐条核对这 9 项

Compatibility兼容性

  • DSH, Node, OS and profile requirementsDSH 版本 / Node 版本 / 操作系统 / profile 是否满足要求
  • External dependencies and runtimes (Electron / Python / Docker, ...)外部依赖与运行时(Electron / Python / Docker 等)是否齐备
  • Conflicts with installed plugins: command names, skill / tool names, ports, duplicate MCP registration与已装插件是否冲突:命令名、skill / tool 重名、端口占用、重复 MCP 注册

Security安全性

  • Repo matches the facts registered here; archived or abandoned?仓库是否与页面登记一致,是否归档或长期停更
  • Safety of preinstall / install / postinstall and install.sh / setup.ps1preinstall / install / postinstall 与 install.sh、setup.ps1 是否安全
  • curl|bash, download-then-execute, obfuscation, unrelated domains → stop immediatelycurl|bash、下载即执行、混淆代码、无关域名 → 立刻停止
  • Typosquatting or unmaintained packages among the new dependencies新增依赖里有没有 typosquatting 或无人维护的包
  • Requested permissions vs. what the feature actually needs申请了哪些权限、是否超出功能所需(filesystem / network / shell / clipboard)
  • Any sudo / admin requirement, plus uninstall and rollback是否要求 sudo / 管理员权限,以及卸载与回滚方式

Anything uncertain must be marked unknown with a note on how to confirm it. This site's signal screen is a static snapshot, not a security audit.拿不准的必须标「未知」并说明要我怎么确认。本站的信号筛查是静态快照,不能替代安全审计。

Or use CLI install (for developers)或使用命令行安装(适合开发者)

CLI Install命令行安装

dsh plugin --profile web add dsh-ci-doctor

把 jkrandom-sudo/dsh-ci-doctor 加入你的 DSH 配置(web profile)即可启用。

READMEREADME

dsh-ci-doctor

中文 · npm · DeepSeek Harness

CI failure, diagnosed before you even open the logs. dsh-ci-doctor watches GitHub Actions for new failures, turns raw job logs into structured findings — normalized error signatures, failure categories, suspect files, trimmed log excerpts — and remembers every signature it has ever seen, so repeat failures are called out on sight. All through two agent tools and the gh CLI you already have authenticated.

Usage

Just ask your agent in plain language — it picks the right tool:

  • "Watch CI on this repo and tell me when something fails" → starts a ci_watch background job.
  • "Why did the nightly build fail?" → runs ci_diagnose on the latest failed run and hands you the diagnosis card.
  • "Diagnose run 31782742089 on cli/cli" → targeted diagnosis of one specific run.

You get a markdown diagnosis card straight in the chat:

## CI diagnosis: cli/cli run #31782742089

**Conclusion:** failure · [run](https://github.com/cli/cli/actions/runs/31782742089)

### Job: Issue Triage (skills-driven)

**Failed steps:** triage
**Signatures:**

- `81a0edf32878` (timeout, first time seen) — server:http_server Session timeout configured…
  **Suspect files:** `script/triage.ts`

<details><summary>Log excerpt</summary>
…
</details>

What it does

1. Watch — the ci_watch tool. Starts a background job that polls for new failed runs (the first poll sets a baseline, so historical red runs never fire the alarm):

{ "repo": "owner/name", "branch": "main", "intervalSeconds": 30, "timeoutMinutes": 60 }
  • Streams status lines you can read any time; cancel any time.
  • Backs off exponentially on transient errors, gives up after 5 consecutive failures, fails immediately on auth errors.
  • On detection it settles with a ready-made next step: call ci_diagnose with repo="…" runId=….
  • Watch a repo explicitly, or omit repo to watch the current working directory's repository.

2. Diagnose — the ci_diagnose tool. Point it at a run (or the latest failed run) and it returns the card shown above:

Showing the opening section of the README — the full document lives in the repository以上为 README 开头摘要,完整文档在仓库内 · View the full README on GitHub →在 GitHub 查看完整 README →

← 上一个 Prev dsh-invoice-tools 下一个 Next dsh-context-console →