agentleFS
alex-voloshin-devalex-voloshin-dev/ai-skills/.agents/skills/deploy-production/SKILL.md

deploy-production

Skill2 starsChanged 6 months agoMITView on GitHub

Deploy to production workflow — final checks, approval gate, deploy, verify, rollback plan. Uses the `deployment-procedures` skill. Requires explicit APPROVE before any production mutation.

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

14 rules. About 0 are common, 12 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈12

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read AGENTS.md at the project root to identify:

    line 16
  2. Very specific:

    If staging was not verified — STOP.

    line 35
  3. Very specific:

    If Risk = HIGH, apply Agent(sre-engineer) for SLO impact assessment.

    line 58
  4. Distinctive:

    Only after the user explicitly approves:

    line 98
  5. Distinctive:

    Verify migration completed successfully before proceeding.

    line 107
  6. Distinctive:

    If using canary deployment:

    line 126
  7. Distinctive:

    If metrics are healthy → proceed to full rollout

    line 129
  8. Distinctive:

    If degradation detected → rollback canary immediately

    line 130
  9. Distinctive:

    Error rate: Should not increase vs pre-deploy baseline

    line 156
  10. Distinctive:

    CPU/Memory: No unusual spikes

    line 158
  11. Distinctive:

    Logs: No new error patterns

    line 159
  12. Distinctive:

    If issues detected — execute rollback immediately:

    line 173
  13. Distinctive:

    Verify application is healthy

    line 191
  14. Distinctive:

    Create incident ticket for investigation

    line 193

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Deploy to Production

Production deployment with mandatory approval gates, verification, and rollback plan. Every step that mutates production requires explicit user APPROVE.

⚠️ SAFETY: No production mutation runs without explicit user APPROVE.

0. Gather Context

Read AGENTS.md at the project root to identify:

  • Cloud platform (GCP, Azure, AWS)
  • Deployment method (Kubernetes, Helm, Docker Compose, serverless)
  • Production environment configuration (namespace, cluster, region)
  • CI/CD pipeline (GitHub Actions, GitLab CI, Jenkins)

This determines which deployment commands and health checks apply.

1. Pre-Deployment Checklist

1a. Staging Verification

Confirm that staging deployment was successful:

  • /deploy-staging completed successfully
  • QA testing passed on staging
  • No critical bugs found in staging
  • Performance acceptable on staging

If staging was not verified — STOP. Run /deploy-staging first.

1b. Release Readiness

  • Version tagged (/release completed)
  • Changelog updated
  • All tests pass on the release branch
  • Database migrations tested (forward and rollback)
  • Feature flags configured (if applicable)
  • Monitoring dashboards ready
  • On-call team notified

1c. Risk Assessment

FactorAssessment
Breaking changesYes/No — migration guide ready?
Database migrationsYes/No — reversible?
Infrastructure changesYes/No — /infra-change completed?
Third-party dependenciesYes/No — API compatibility verified?
Traffic impactLow/Medium/High
Rollback complexitySimple (revert image) / Complex (DB migration)

If Risk = HIGH, apply Agent(sre-engineer) for SLO impact assessment.

2. Prepare Deployment

2a. Identify Deployment Method

Same as /deploy-staging Step 1c — but with production configuration.

2b. Backup Current State

# Record current deployment state for rollback
kubectl get deployment -n production -o yaml > pre-deploy-state.yaml

Or for Helm:

helm get values <release> -n production > pre-deploy-values.yaml
helm history <release> -n production

2c. Deployment Plan

Present the deployment plan:

## Production Deployment Plan
- **Version**: [current] → [new]
- **Method**: [K8s/Helm/Docker/Platform]
- **Migrations**: [list if any]
- **Config changes**: [list if any]
- **Expected downtime**: [none / X minutes]
- **Rollback plan**: [documented in Step 5]
- **Monitoring**: [dashboards to watch]

⚠️ STOP. Request APPROVE before proceeding to Step 3.

3. Deploy — REQUIRES EXPLICIT APPROVE

Only after the user explicitly approves:

3a. Run Migrations (if applicable)

# Database migration
<migration-command>

Verify migration completed successfully before proceeding.

3b. Deploy Application

Kubernetes / Helm:

helm upgrade --install <release> <chart> \
  -n production \
  -f values-production.yaml \
  --set image.tag=<tag>

Rolling update strategy — monitor pod rollout:

kubectl rollout status deployment/<name> -n production --timeout=300s

3c. Canary (if applicable)

If using canary deployment:

  1. Deploy to canary (small % of traffic)
  2. Monitor error rates and latency for 5–10 minutes
  3. If metrics are healthy → proceed to full rollout
  4. If degradation detected → rollback canary immediately

4. Post-Deployment Verification

4a. Health Check

// turbo
kubectl get pods -n production -o wide
curl -s <production-url>/health

4b. Smoke Tests

  • Application responds on production URL
  • Health endpoint returns 200
  • Authentication flow works
  • Core API endpoints respond correctly
  • Key user journeys functional

4c. Monitor (5–10 minutes)

Watch for:

  • Error rate: Should not increase vs pre-deploy baseline
  • Latency: P50, P95, P99 within SLO targets
  • CPU/Memory: No unusual spikes
  • Logs: No new error patterns
// turbo
kubectl logs -n production -l app=<app-name> --tail=100 --since=5m
// turbo
kubectl get events -n production --sort-by='.lastTimestamp' --field-selector type!=Normal

5. Rollback Plan

If issues detected — execute rollback immediately:

Helm:

helm rollback <release> <previous-revision> -n production

Kubernetes:

kubectl rollout undo deployment/<name> -n production

Database migration rollback:

<rollback-migration-command>

After rollback:

  • Verify application is healthy
  • Notify team of rollback
  • Create incident ticket for investigation

6. Summary

## Production Deployment Summary
- **Version**: [old] → [new]
- **Deployed at**: [timestamp]
- **Method**: [K8s/Helm/Docker/Platform]
- **Migrations**: [applied/N/A]
- **Health check**: [pass/fail]
- **Smoke tests**: [pass/fail]
- **Monitoring**: [stable/issues detected]
- **Rollback**: [not needed / executed at timestamp]
- **Production URL**: [url]
- **Next steps**: [monitoring period, team notification, release announcement]

Integration

  • Preceded by: /deploy-staging (staging verification), /release (version tag)
  • Roles: Agent(devops-engineer), Agent(sre-engineer), Agent(devops-architect) (deployment strategy design)
  • Skills: deployment-procedures skill

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.