Cheggin/post-deploy-loop

Post-Deploy Loop

After deploy, the harness doesn't stop. This skill runs the ongoing phases.

The Loop

Every tick (configurable, default 1 hour):

1. Health check

• Fetch production URL — is it responding?
• Check Sentry MCP for new errors since last tick
• Check uptime status

If unhealthy: dispatch ops agent → diagnose → fix → deploy → verify.

2. Metrics check

• Query PostHog (if configured) for key metrics
• Compare against previous tick: traffic, conversion, errors
• Flag significant changes (>20% swing either direction)

3. Content queue

• Check if blog posts are scheduled
• Check social media queue
• If content is due: dispatch writing agent

4. Dependency check (weekly)

• Run npm audit
• Check for outdated packages
• Create GitHub Issues for critical vulnerabilities

5. Investor update (weekly)

• Aggregate: features shipped, traffic, errors, revenue (if Stripe), user count
• Post to Slack

6. Feature iteration

• Check GitHub Issues for user feedback / feature requests
• Prioritize by votes/impact
• If P0 feature request: add to build queue

Tick Frequency

| Check | Frequency | |-------|-----------| | Health | Every tick (1h) | | Metrics | Every tick (1h) | | Content | Daily | | Dependencies | Weekly | | Investor update | Weekly | | Feature iteration | On new feedback |

Stop Conditions

• User says stop
• Production URL deleted/unreachable for 24h+
• Budget exceeded