Bobby Watchdog

---
name: watchdog-check
description: "Post-Deploy Watchdog Skill: Verifies production health after deployment. Checks page loads, console errors, and API responses. MANDATORY TRIGGERS: watchdog, post-deploy, verify deploy, production check, health check, smoke test."
argument-hint: ""
---

# Bobby Watchdog Skill

> Post-deploy verification. Checks that production pages load, APIs respond, and no new console errors appear after deployment.

## Before Starting

1. **Check learnings** — Read `.claude/skills/bobby-watchdog/learnings.md` + `.claude/skills/bobby-watchdog/learnings.local.md`
2. Read `.bobbyrc.yml` for production config:
   - `production_url` — base URL for production
   - `watchdog_pages` — list of paths to check (default: `["/"]`)

## Watchdog Check Process

### Step 1: Determine What to Check

Check these pages (in order):
1. Pages listed in `watchdog_pages` config
2. If no config: homepage (`/`), login page, main dashboard
3. Any pages specifically affected by the most recent deployment

### Step 2: Health Checks

For each page, verify via browser automation:

**Page Load:**
- [ ] HTTP status is 200 (or expected redirect)
- [ ] Page renders within 5 seconds
- [ ] No blank/white page (check that main content container exists)
- [ ] Title tag is present and not "Error" or "500"

**Console Errors:**
- [ ] No JavaScript errors in console
- [ ] No failed network requests (4xx/5xx) except expected ones
- [ ] No unhandled promise rejections

**Visual Verification:**
- [ ] Screenshot each page for the record
- [ ] Key UI elements are present (navigation, main content, footer)
- [ ] No "loading" spinners stuck indefinitely

**API Health (if applicable):**
- [ ] Health endpoint responds: `curl -s {production_url}/api/health`
- [ ] Key API endpoints return expected status codes

### Step 3: Compare to Pre-Deploy State

If previous watchdog results exist (`.bobby/watchdog/last-run.json`):
- Compare screenshot diffs for unexpected visual changes
- Compare console error counts (new errors = regression)
- Compare response times (>50% slower = regression)

### Step 4: Record Results

Save to `.bobby/watchdog/run-{YYYYMMDD-HHmmss}.json`:

```json
{
  "timestamp": "ISO date",
  "production_url": "https://...",
  "pages": [
    {
      "path": "/",
      "status": 200,
      "load_time_ms": 850,
      "console_errors": 0,
      "failed_requests": 0,
      "screenshot": "watchdog/screenshots/home-{timestamp}.png",
      "passed": true
    }
  ],
  "overall": "passed"
}
```

Also save as `last-run.json` for future comparisons.

## Decision

### All Checks Pass
Output:
```
Watchdog check PASSED — {timestamp}
Production: {url}
Pages checked: {n}
Console errors: 0
Failed requests: 0
All pages loaded within 5s.
```

### Issues Found
1. File bug tickets for each issue:
   ```bash
   bobby ticket create -t "WATCHDOG: {page} returns {status}" --type bug -p critical
   bobby ticket create -t "WATCHDOG: JS error on {page}: {error message}" --type bug -p high
   ```
2. Output:
```
Watchdog check FAILED — {timestamp}
Production: {url}
Issues found:
- {page}: {issue description}
- {page}: {issue description}
Tickets filed: {ticket IDs}
```

## If Something Goes Wrong

- Production URL not configured: Ask user to add `production_url` to `.bobbyrc.yml`
- Production is completely down: `bobby ticket create -t "WATCHDOG: Production is DOWN — {url} unreachable" --type bug -p critical`
- If you discovered a deployment pattern: `bobby learn bobby-watchdog "pattern" "description"`

---

## Project overrides

If `.claude/skills/bobby-watchdog/SKILL.local.md` exists, read it and follow it. It holds this
project's own instructions for this skill and **wins** wherever it conflicts with anything
above.

`SKILL.md` is shipped by Bobby and is replaced on every upgrade — edits here are lost.
`SKILL.local.md` is yours and is never overwritten.