Deployment Verification

Closes the loop on deployments by verifying they completed successfully.

When to Use

Trigger /verify-deploy after:

• Pushing changes to master
• Waiting ~2-3 minutes for auto-deploy
• Want to confirm deployment succeeded

Don't use if:

• Haven't pushed yet (use build verification instead)
• Just pushed seconds ago (deployments take 2-3 minutes)
• Making local changes only

What This Checks

For Dashboard Changes (Vercel)

1. Build Status

   • Check latest Vercel deployment
   • Verify build completed without errors
   • Get deployment URL

2. Deployment Health

   • Hit health endpoint (if exists)
   • Verify returns 200 OK
   • Check response time

3. Key Pages Accessible

   • Test critical routes load
   • No 500 errors
   • No obvious visual breaks

For Pipeline Changes (Cloud Run)

1. Build Status

   • Check latest Cloud Build
   • Verify Docker build succeeded
   • Confirm deployment to Cloud Run

2. Service Health

   • Hit /health endpoint
   • Verify Supabase connection
   • Check response structure

3. Recent Logs

   • Check for startup errors
   • Verify no crashes
   • Confirm service is stable

Execution Flow

Step 1: Identify What Was Deployed

Check git diff:

git log -1 --name-only

Determine:

• Dashboard changes? (files in dashboard/)
• Pipeline changes? (files in src/feedops/)
• Both?

Step 2: Vercel Verification (If Dashboard Changes)

Option A: Using Vercel MCP (if available)

// Use mcp__vercel__ tools to check deployment status
// Get latest deployment
// Check build status
// Get deployment URL

Option B: Using Vercel CLI

# Check latest deployment
vercel list --project-id=prj_00zlLdZVgbP8XjDWIEXSRdFyqDqA --limit=1

# Get deployment status
vercel inspect [deployment-url]

Option C: Check Vercel dashboard logs

# View logs via Vercel API or dashboard link
echo "View logs: https://vercel.com/[team]/[project]/deployments"

What to verify:

• ✅ Status: READY (not ERROR or BUILDING)
• ✅ Build time: Reasonable (~1-2 minutes for this project)
• ✅ Commit hash: Matches latest push
• ⚠️ Warnings: Check for deprecation notices

Test health:

# Hit production URL
curl -I https://allied-feed-ops.vercel.app/api/health

# Expected: 200 OK

Step 3: Cloud Run Verification (If Pipeline Changes)

Check Cloud Build:

gcloud builds list \
  --project=bobbys-project-346400 \
  --limit=1 \
  --format="table(id,status,createTime,duration)"

What to verify:

• ✅ Status: SUCCESS
• ✅ Create time: Recent (within last 5 minutes)
• ✅ Duration: Reasonable (~3-5 minutes for this project)

Check Cloud Run deployment:

gcloud run services describe feedops-pipeline \
  --project=bobbys-project-346400 \
  --region=us-east1 \
  --format="value(status.latestReadyRevisionName,status.conditions)"

What to verify:

• ✅ Latest revision matches recent deployment
• ✅ Conditions: All READY
• ⚠️ Check for any warnings

Test health endpoint:

curl -s https://feedops-pipeline-623866089882.us-east1.run.app/health | jq .

# Expected response:
# {
#   "status": "healthy",
#   "supabase_connected": true,
#   "timestamp": "[recent-time]"
# }

Check recent logs:

gcloud run services logs read feedops-pipeline \
  --project=bobbys-project-346400 \
  --limit=20 \
  --format="table(timestamp,severity,textPayload)"

# Look for:
# - ✅ "Application startup complete"
# - ✅ No ERROR severity logs
# - ⚠️ Any unexpected warnings

Step 4: Report Results

Format:

## Deployment Verification Results

**Commit:** [hash] - [message first line]
**Pushed:** [timestamp]
**Verified:** [current timestamp]

### Dashboard (Vercel)
- Status: [✅ DEPLOYED / ⚠️ BUILDING / ❌ FAILED]
- Build: [SUCCESS / ERRORS]
- URL: [deployment URL]
- Health: [✅ 200 OK / ❌ Error]
[Details if issues]

### Pipeline (Cloud Run)
- Status: [✅ DEPLOYED / ⚠️ BUILDING / ❌ FAILED]
- Build: [SUCCESS / ERRORS]
- Service: [READY / NOT READY]
- Health: [✅ Healthy / ❌ Unhealthy]
[Details if issues]

### Overall: [✅ ALL SYSTEMS GO / ⚠️ WARNINGS / ❌ FAILURES]

If all successful:

✅ Deployment Verified Successfully

All systems operational:
- Dashboard: https://allied-feed-ops.vercel.app
- Pipeline: https://feedops-pipeline-623866089882.us-east1.run.app

Changes from commit [hash] are live.

If issues found:

❌ Deployment Issues Detected

Dashboard: [issue description]
- [Details]
- [Logs/errors]

Pipeline: [issue description if applicable]
- [Details]
- [Logs/errors]

Recommended action: [rollback / investigate / wait]

Edge Cases

Still Building

⏳ Deployment In Progress

Dashboard: BUILDING (started [time] ago)
Pipeline: BUILDING (started [time] ago)

Typical deploy time: 2-3 minutes

Recommendation: Wait 2 more minutes, then run /verify-deploy again

Partial Success

⚠️ Partial Deployment

Dashboard: ✅ Deployed successfully
Pipeline: ❌ Build failed with errors:
[error details]

Since dashboard changes succeeded, those are live.
Pipeline issues need investigation.

Rollback Needed

If critical issues detected:

🔴 Critical Issues - Rollback Recommended

Issue: [description of what's broken]
Impact: [production / high / medium]

Rollback options:

1. **Instant (Vercel):** Revert to previous deployment in Vercel dashboard
2. **Git revert:** `git revert [commit-hash] && git push`
3. **Wait for fix:** If low impact, fix forward

Recommend: [Option number] because [reason]

Should I execute rollback, or investigate further?

Integration with /wrapup

Enhanced wrapup flow:

1. /wrapup executes (CLAUDE.md, build, commit, push)
2. Push completes
3. Auto-suggest: "Wait 2-3 minutes for deploy, then run /verify-deploy"
4. User runs /verify-deploy after waiting
5. Confirmation deployment succeeded

Automation Opportunity

Future enhancement: Hook that runs automatically 3 minutes after push:

{
  "postToolUse": [
    {
      "tool": "Bash",
      "pattern": ".*git push.*",
      "command": "bash -c 'echo \"⏰ Deploy verification: Run /verify-deploy in 2-3 minutes\"'"
    }
  ]
}

Time Estimate

Typical /verify-deploy execution: 30-60 seconds

• Cloud Build check: 5s
• Cloud Run check: 5s
• Health endpoints: 5s
• Vercel check: 5-10s
• Log review: 10-20s
• Report generation: 5s