devops-bridge

# 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:

```
🔴 CI Failed — PR #142 "Add OAuth flow" by @alice
  └─ Test: auth.test.ts:47 — Expected 200, got 401
  └─ Last passing commit: abc1234 (2 hours ago)  
  └─ Linked issue: LINEAR-389 "Implement SSO"
  └─ Action: Reply "fix" to see the failing test, "logs" for full output
```

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:
  ```
  👀 Review needed:
  • PR #142 "Add OAuth flow" — waiting 3 days (assigned: @bob)
  • PR #156 "Fix pagination" — waiting 1 day (assigned: @carol)
  ```

**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:

```
🧑‍💻 Dev Standup — [Date]

Merged yesterday:
  • PR #140 "Refactor auth module" by @alice → LINEAR-385 closed
  • PR #143 "Update deps" by @bob

In review:
  • PR #142 "Add OAuth flow" by @alice — 2 approvals, CI passing ✅
  • PR #156 "Fix pagination" by @carol — changes requested by @bob

Blocked:
  • PR #158 "Migrate DB" by @dave — CI failing (migration timeout)
  • Issue LINEAR-402 — no assignee, due tomorrow

CI Health: 87% pass rate (down from 94% last week)
  └─ Flaky: auth.test.ts (failed 4/10 runs)
```

## 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 `gh`

### 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"
2. **Where to send notifications?** Slack channel, Discord channel, Telegram, or all
3. **How aggressive should reminders be?** Options: gentle (72h), moderate (48h), aggressive (24h)
4. **Include CI details?** Some users want full logs, others just pass/fail
5. **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):

```json
[
  {
    "name": "Morning dev digest",
    "schedule": "0 9 * * 1-5",
    "prompt": "Generate dev standup summary for my repos"
  },
  {
    "name": "PR review reminder",  
    "schedule": "0 14 * * 1-5",
    "prompt": "Check for PRs waiting for review and send reminders"
  },
  {
    "name": "End of day CI report",
    "schedule": "0 17 * * 1-5",
    "prompt": "Summarize today's CI/CD activity and flag any issues"
  }
]
```

## 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