run-deploy

Goal

Deploy the application to the target platform with post-deploy validation, commit/push, and browser watching until the deployment is live

Scope

• For related skills: open-web-for-config-secret, deep-review

Covers verify, build, deploy, commit/push, post-deploy validation, watch until live, and rollback & recovery

Execute

Pre-Run: Always run /deep-review first — run-* requires review/assessment before the main action, never skip; if findings are blockers, fix or report them before running (deploy)

1. Verify

Goal: Verify

Check code quality before deploy

1. Run /run-verify to check code quality
2. Confirm verify succeeds
3. Confirm there are no significant errors or warnings

2. Build Application

Goal: Build Application

Build the application for deployment

1. Run /run-build to build the application
2. Confirm the build succeeds
3. Inspect the created artifacts

3. Deploy

Goal: Deploy

Deploy the application per the platform in use

1. For Vercel: run /deploy-to-vercel
2. For Cloudflare:
   • If the Worker project has not been created yet → run /create-cloudflare-worker first
   • Then run /deploy-to-cloudflare
3. For Railway: run /deploy-to-railway
4. For other platforms: follow the relevant workflow
5. Confirm the deploy succeeds

4. Commit And Push

Goal: Commit And Push

Commit and push changes to the repository

1. Run /git-commit
2. Run /git-push

5. Verify Deployment

Goal: Verify Deployment

Confirm the deployment works

1. Open the deployment URL
2. Confirm the application works
3. Check the logs for errors
4. Run /watch-browser fix if there is a URL

6. Watch Until Live

Goal: Watch Until Live

Watch the deployment with a browser until it is live

1. Run /watch-browser fix with the deployment URL
2. Confirm the page loads successfully
3. Check console errors and network errors
4. Run /resolve-errors if problems are found
5. Run /loop-until-complete until the deployment is successfully live

7. Post-Deploy Validation

Goal: Post-Deploy Validation

Thoroughly inspect the deployment after deploy (see references/post-deploy-validation.md)

1. Run /run-test (e2e) to test critical paths after deploy
2. Check the health endpoint and readiness probes
3. Check error logs for 5-10 minutes after deploy
4. If critical errors are found → roll back immediately (see references/rollback-recovery.md)

8. Rollback And Recovery

Goal: Rollback And Recovery

Prepare the rollback procedure and check the backup (see references/rollback-recovery.md)

1. Run /deep-review to check the backup strategy before deploy
2. Prepare a clear rollback procedure before deploy
3. If the deploy fails → roll back immediately and run /resolve-errors

Rules

1. Verify

• Verify must succeed before build
• Use /run-verify for verify
• No significant errors or warnings

2. Build

• Build must succeed before deploy
• Artifacts must be correct
• Use /run-build for build

3. Deploy

• Use /follow-secret-manager to manage secrets and API tokens before deploy
• Use /deploy-to-<platform> per the chosen platform
• Deploy must succeed
• Check deployment status
• If deploying to Cloudflare and the deploy fails, run /resolve-cicd to watch and fix until it passes

4. Commit And Push

• Deploy before commit/push if needed
• Commit/push after the deploy succeeds
• Watch the browser once the deployment URL is available

5. Watch Until Live

• Use /watch-browser fix for monitoring
• Use /resolve-errors when problems are found
• Check console and network errors
• Run /loop-until-complete until the deployment is live

6. Post-Deploy Validation

• Critical paths must be tested after deploy
• The health endpoint must be checked
• Error logs must be checked for 5-10 minutes after deploy
• If critical errors are found, roll back immediately

7. Rollback And Recovery

• The rollback procedure must be prepared before deploy

• The backup strategy must be checked before deploy

• If the deploy fails, roll back immediately

• Zero-downtime deploy is required (when needed)

• Use /follow-tool-vite if needed

• Use /follow-create-plugins (vite) if needed (deploy)

• Use /deep-test smoke in Post-Deploy Validation — sanity check critical paths on the live URL

Expected Outcome

• Application deployed successfully
• Changes committed and pushed
• URL is accessible
• Application works normally
• No console or network errors
• Post-deploy validation passes (smoke test, health check, logs)
• Rollback procedure is ready to use
• Backup strategy checked before deploy