DevOps Bridge
The missing link between your dev tools. This skill connects GitHub, CI/CD, Slack/Discord, and issue trackers into workflows that actually make sense — so you stop context-switching between 6 browser tabs.
Why This Exists
Developers already use separate skills for GitHub, Slack, etc. But nobody has built the bridge: when CI fails, automatically link it to the PR, notify the right Slack channel, and update the ticket. This skill is that bridge.
Core Capabilities
1. Smart Notifications
Transform noisy GitHub events into actionable, context-rich messages. Instead of "Build failed", deliver:
CODEBLOCK0
When sending notifications, always:
- - Include the PR title and author, not just the number
- Link to the specific failing test or check, not just "CI failed"
- Mention the last known good commit for quick bisect context
- Cross-reference related issues/tickets if they exist
- Suggest a concrete next action
2. PR Review Management
Track pull request lifecycle across tools:
Review reminders:
- - Scan open PRs daily and flag those waiting for review
- Escalate based on age: gentle reminder at 24h, stronger at 48h, urgent at 72h+
- Send reminders to the assigned reviewer via Slack/Discord DM or channel
- Format:
CODEBLOCK1
Review status sync:
- - When a PR gets approved on GitHub, post to the team channel
- When changes are requested, notify the author directly
- When all checks pass + approved, prompt: "Ready to merge — want me to merge it?"
3. CI/CD Intelligence
Go beyond "pass/fail" with intelligent CI analysis:
- - Failure grouping: if multiple PRs fail on the same test, flag it as a systemic issue rather than spamming individual notifications
- Flaky test detection: if a test fails intermittently across PRs, note it: "This test has failed 3 times this week across different PRs — likely flaky"
- Duration tracking: "This build took 45 min, up from the usual 20 min — something may be wrong"
- Auto-retry suggestion: for known flaky failures, suggest or trigger a re-run
4. Issue Tracker Sync
Keep issue trackers (Linear, Jira, GitHub Issues) in sync with actual development activity:
- - When a PR references an issue (e.g., "Fixes #123"), update the issue status automatically
- When a PR is merged, move the linked issue to "Done" or "In Review"
- When CI fails on a PR linked to an issue, add a comment to the issue noting the blocker
- Surface orphaned PRs: "PR #167 doesn't reference any issue — should it?"
5. Daily Dev Standup
Generate a team-level development summary on demand or via cron:
CODEBLOCK2
Configuration
Required Tools
- -
gh CLI (GitHub) — for repo activity, PRs, issues, CI status - At least one messaging channel configured (Slack, Discord, Telegram)
Optional Tools
- - Linear CLI or API — for Linear issue tracking
- Jira API — for Jira integration
- GitHub Issues — works out of the box with INLINECODE1
Setup Flow
On first use, gather configuration interactively:
- 1. Which repos to monitor? Ask for a list or use "all repos I have push access to"
- Where to send notifications? Slack channel, Discord channel, Telegram, or all
- How aggressive should reminders be? Options: gentle (72h), moderate (48h), aggressive (24h)
- Include CI details? Some users want full logs, others just pass/fail
- Who's on the team? Map GitHub usernames to Slack/Discord handles for @mentions
Store configuration in workspace memory for persistence.
Cron Setup
Suggest these default schedules (user can customize):
CODEBLOCK3
Command Reference
Users can trigger specific actions with natural language:
| User says | Action |
|---|
| "What's happening in my repos?" | Full activity summary across all monitored repos |
| "Any failing builds?" |
CI status check with details on failures |
| "Who needs a review?" | List PRs awaiting review with age and assignee |
| "Standup" | Generate daily dev standup summary |
| "Notify #dev-team about PR 142" | Send a formatted notification about a specific PR |
| "Link PR 142 to LINEAR-389" | Create cross-reference between PR and issue |
| "Set up CI alerts for repo X" | Configure monitoring for a specific repository |
| "Merge PR 142" | Merge if all checks pass and approved; warn if not |
Edge Cases
- - Monorepo: if monitoring a monorepo, group notifications by directory/team, not just by PR
- Forks: when PRs come from forks, note this clearly (different trust level)
- Draft PRs: don't send review reminders for draft PRs unless user asks
- Stale PRs: if a PR has been open >7 days with no activity, suggest closing or rebasing
- Rate limits: GitHub API has rate limits. Batch requests and cache results within a session
- No messaging configured: if no Slack/Discord/Telegram, output to the current conversation instead
- Multiple orgs: if user has repos across multiple GitHub orgs, handle them all but label clearly
Integration Pattern
This skill works best with other installed skills. When detected:
- - github skill: defer raw GitHub operations to it, use devops-bridge for cross-tool logic
- slack skill: use it for message delivery, devops-bridge composes the messages
- daily-briefing-hub: feed dev activity data into the morning briefing's "Dev Activity" section
DevOps Bridge
连接开发工具的关键纽带。该技能将GitHub、CI/CD、Slack/Discord和问题追踪器整合为真正有意义的工作流——让你无需在6个浏览器标签页之间频繁切换。
为何存在
开发者已拥有GitHub、Slack等独立技能,但无人构建桥梁:当CI失败时,自动关联PR、通知对应Slack频道并更新工单。本技能正是这座桥梁。
核心能力
1. 智能通知
将嘈杂的GitHub事件转化为可操作、富含上下文的消息。不再显示构建失败,而是呈现:
🔴 CI失败 — PR #142 添加OAuth流程 由 @alice 提交
└─ 测试: auth.test.ts:47 — 期望200,实际返回401
└─ 上次通过提交: abc1234 (2小时前)
└─ 关联问题: LINEAR-389 实现SSO
└─ 操作: 回复fix查看失败测试,logs查看完整输出
发送通知时,始终:
- - 包含PR标题和作者,而非仅编号
- 链接到具体失败的测试或检查项,而非仅CI失败
- 提及上次已知的良好提交,便于快速二分定位
- 交叉引用相关问题/工单(如存在)
- 建议具体的下一步操作
2. PR审查管理
跨工具追踪拉取请求生命周期:
审查提醒:
- - 每日扫描开放PR,标记等待审查的请求
- 按时间升级提醒:24小时温和提醒,48小时加强提醒,72小时以上紧急提醒
- 通过Slack/Discord私信或频道发送给指定审查者
- 格式:
👀 需要审查:
• PR #142 添加OAuth流程 — 等待3天(指定审查者:@bob)
• PR #156 修复分页 — 等待1天(指定审查者:@carol)
审查状态同步:
- - 当PR在GitHub上获得批准时,发布到团队频道
- 当要求修改时,直接通知作者
- 当所有检查通过且获得批准时,提示:准备合并——需要我合并吗?
3. CI/CD智能分析
超越简单的通过/失败,提供智能CI分析:
- - 失败分组:如果多个PR在相同测试上失败,标记为系统性问题,而非发送大量单独通知
- 不稳定测试检测:如果某个测试在不同PR间间歇性失败,注明:该测试本周在不同PR上已失败3次——可能不稳定
- 时长追踪:本次构建耗时45分钟,远超通常的20分钟——可能存在问题
- 自动重试建议:对于已知的不稳定失败,建议或触发重新运行
4. 问题追踪器同步
保持问题追踪器(Linear、Jira、GitHub Issues)与实际开发活动同步:
- - 当PR引用某个问题时(例如修复 #123),自动更新问题状态
- 当PR被合并时,将关联问题移至已完成或审查中
- 当与问题关联的PR上CI失败时,在问题中添加评论说明阻塞原因
- 显示孤立PR:PR #167 未引用任何问题——是否需要关联?
5. 每日开发站会
按需或通过定时任务生成团队级开发摘要:
🧑💻 开发站会 — [日期]
昨日合并:
• PR #140 重构认证模块 由 @alice 提交 → LINEAR-385 已关闭
• PR #143 更新依赖 由 @bob 提交
审查中:
• PR #142 添加OAuth流程 由 @alice 提交 — 2个批准,CI通过 ✅
• PR #156 修复分页 由 @carol 提交 — @bob 要求修改
阻塞中:
• PR #158 迁移数据库 由 @dave 提交 — CI失败(迁移超时)
• 问题 LINEAR-402 — 无负责人,明天截止
CI健康度:87%通过率(上周为94%)
└─ 不稳定:auth.test.ts(10次运行失败4次)
配置
必需工具
- - gh CLI(GitHub)— 用于仓库活动、PR、问题、CI状态
- 至少配置一个消息通道(Slack、Discord、Telegram)
可选工具
- - Linear CLI或API — 用于Linear问题追踪
- Jira API — 用于Jira集成
- GitHub Issues — 与gh开箱即用
设置流程
首次使用时,交互式收集配置:
- 1. 监控哪些仓库? 询问列表或使用我有推送权限的所有仓库
- 通知发送到哪里? Slack频道、Discord频道、Telegram或全部
- 提醒强度如何? 选项:温和(72小时)、适中(48小时)、激进(24小时)
- 是否包含CI详情? 部分用户需要完整日志,其他仅需通过/失败
- 团队成员有哪些? 将GitHub用户名映射到Slack/Discord句柄以便@提及
将配置存储在工作区内存中以保持持久性。
定时任务设置
建议以下默认计划(用户可自定义):
json
[
{
name: 早间开发摘要,
schedule: 0 9 1-5,
prompt: 为我的仓库生成开发站会摘要
},
{
name: PR审查提醒,
schedule: 0 14 1-5,
prompt: 检查等待审查的PR并发送提醒
},
{
name: 日终CI报告,
schedule: 0 17 1-5,
prompt: 总结今日CI/CD活动并标记任何问题
}
]
命令参考
用户可通过自然语言触发特定操作:
| 用户说 | 操作 |
|---|
| 我的仓库发生了什么? | 所有监控仓库的完整活动摘要 |
| 有构建失败吗? |
CI状态检查,包含失败详情 |
| 谁需要审查? | 列出待审查的PR及其等待时间和审查者 |
| 站会 | 生成每日开发站会摘要 |
| 通知 #dev-team 关于PR 142 | 发送关于特定PR的格式化通知 |
| 将PR 142关联到LINEAR-389 | 创建PR与问题之间的交叉引用 |
| 为仓库X设置CI告警 | 配置特定仓库的监控 |
| 合并PR 142 | 如果所有检查通过且已批准则合并;否则发出警告 |
边界情况
- - 单仓库:如果监控的是单仓库,按目录/团队分组通知,而非仅按PR
- 分支仓库:当PR来自分支仓库时,明确标注(不同信任级别)
- 草稿PR:除非用户要求,否则不发送草稿PR的审查提醒
- 过期PR:如果PR已开放超过7天且无活动,建议关闭或变基
- 速率限制:GitHub API有速率限制。在会话内批量请求并缓存结果
- 未配置消息通道:如果未配置Slack/Discord/Telegram,则输出到当前对话
- 多个组织:如果用户在多个GitHub组织中有仓库,全部处理但清晰标注
集成模式
该技能与其他已安装技能配合最佳。检测到时:
- - github技能:将原始GitHub操作委托给它,使用devops-bridge处理跨工具逻辑
- slack技能:用于消息投递,devops-bridge负责消息编排
- daily-briefing-hub:将开发活动数据输入早间简报的开发活动部分