Skip to content

mattgierhart/PRD-driven-context-engineering / .claude/skills/prd-v08-release-planning

prd-v08-release-planning

Verified

Define release criteria, deployment environments, and rollback strategies during PRD v0.8 Deployment & Ops. Triggers on requests to plan releases, define deployment criteria, or when user asks "how do we deploy?", "release criteria", "deployment plan", "rollback strategy", "go-live checklist". Outputs DEP- entries with deployment steps and release criteria.

FileKindSize
SKILL.mdskill md12.0 KB
assets/dep-template.mdmarkdown11.8 KB
references/deployment-strategies.mdmarkdown16.5 KB
references/examples.mdmarkdown16.8 KB
references/rollback-procedures.mdmarkdown18.2 KB

assets/dep-template.md

View on GitHub
# DEP-XXX: [Deployment Title]

**ID**: DEP-XXX
**Type**: [Release | Hotfix | Rollback | Migration]
**Environment**: [Staging | Production | Canary | All]
**Strategy**: [Blue-Green | Canary | Rolling | Feature Flag | Big Bang]
**Owner**: [Team/Person responsible for deployment]
**Created**: YYYY-MM-DD
**Target Date**: YYYY-MM-DD
**Status**: [Planned | In Progress | Completed | Rolled Back]

---

## Summary

**What**: [Brief description of what is being deployed]

**Why**: [Business or technical reason for this deployment]

**Risk Level**: [Low | Medium | High | Critical]

**Deployment Window**: [Date/time range - e.g., 2026-01-15 02:00-04:00 UTC]

**Estimated Duration**: [Expected time to complete - e.g., 30 minutes, 2 hours]

**Rollback Estimated Duration**: [Time to rollback if needed - e.g., 5 minutes, 15 minutes]

---

## Changes Included

**Features**:
- [FEA-XXX]: [Feature description]
- [FEA-YYY]: [Feature description]

**Bug Fixes**:
- [Issue #123]: [Bug description]
- [Issue #456]: [Bug description]

**Infrastructure Changes**:
- [ARC-XXX]: [Architecture change description]
- [Infrastructure change description]

**Database Migrations**:
- [DBT-XXX]: [Migration description - e.g., add index, add column, data backfill]

**Dependencies**:
- [External service upgrade - e.g., PostgreSQL 14 → 15]
- [Library upgrade - e.g., React 17 → 18]

---

## Pre-Deployment Checklist

**Code Quality**:
- [ ] All tests passing (TEST-XXX)
- [ ] Code review approved
- [ ] Security scan passed
- [ ] Performance testing completed (if applicable)

**Documentation**:
- [ ] Changelog updated
- [ ] API documentation updated (if API changes)
- [ ] Runbook updated (RUN-XXX)
- [ ] Monitoring updated (MON-XXX) for new metrics/alerts

**Infrastructure**:
- [ ] Infrastructure changes applied to staging
- [ ] Database migrations tested in staging
- [ ] Capacity planning reviewed (traffic projections)
- [ ] DNS/CDN changes prepared (if applicable)

**Communication**:
- [ ] Stakeholders notified (product, support, customers if downtime)
- [ ] On-call schedule confirmed
- [ ] Incident channel prepared (#incidents)
- [ ] Status page updated (if customer-facing downtime expected)

**Rollback Plan**:
- [ ] Rollback procedure documented (see Rollback Plan section)
- [ ] Rollback tested in staging
- [ ] Database rollback plan ready (if schema changes)
- [ ] Feature flags configured (if applicable)

---

## Deployment Steps

### Step 1: Pre-Deployment Validation

**Actions**:
1. [ ] Verify all pre-deployment checklist items complete
2. [ ] Confirm monitoring dashboards accessible (MON-XXX)
3. [ ] Confirm on-call team ready
4. [ ] Take database backup (if schema changes)
5. [ ] Record baseline metrics (request rate, error rate, latency)

**Success Criteria**: All checklist items checked, baselines recorded

**Rollback Trigger**: N/A (not yet deployed)

---

### Step 2: [Deployment Strategy Specific Steps]

**For Blue-Green**:
1. [ ] Deploy to green environment
2. [ ] Run smoke tests on green
3. [ ] Switch 10% traffic to green (canary within blue-green)
4. [ ] Monitor for 15 minutes
5. [ ] Switch 100% traffic to green
6. [ ] Monitor blue environment (keep as rollback target)

**For Canary**:
1. [ ] Deploy to canary nodes (10% of fleet)
2. [ ] Monitor canary metrics for 30 minutes
3. [ ] If healthy, deploy to 50% of fleet
4. [ ] Monitor for 30 minutes
5. [ ] Deploy to 100% of fleet

**For Rolling**:
1. [ ] Deploy to first availability zone (AZ-1)
2. [ ] Monitor AZ-1 for 15 minutes
3. [ ] Deploy to second availability zone (AZ-2)
4. [ ] Monitor AZ-2 for 15 minutes
5. [ ] Deploy to third availability zone (AZ-3)
6. [ ] Monitor full deployment for 30 minutes

**For Feature Flag**:
1. [ ] Deploy code with feature flag OFF
2. [ ] Verify deployment successful (no errors)
3. [ ] Enable feature flag for internal users (10%)
4. [ ] Monitor for 1 hour
5. [ ] Enable for 50% of users
6. [ ] Monitor for 4 hours
7. [ ] Enable for 100% of users

**Success Criteria**: [Specific metrics - e.g., error rate < 0.1%, p95 latency < 500ms, no increase in 5xx]

**Rollback Trigger**: [Specific conditions - e.g., error rate > 1%, p95 latency > 1000ms, 5xx errors > 10/minute]

---

### Step 3: Post-Deployment Validation

**Actions**:
1. [ ] Run smoke tests on production
2. [ ] Verify critical user journeys (UJ-XXX)
   - [ ] [UJ-101]: [Journey name - e.g., User signup]
   - [ ] [UJ-102]: [Journey name - e.g., Checkout flow]
3. [ ] Check monitoring dashboards (MON-XXX)
   - [ ] Request rate within expected range
   - [ ] Error rate < threshold (SLO compliance)
   - [ ] Latency within expected range
4. [ ] Verify database migrations completed (if applicable)
5. [ ] Check logs for unexpected errors
6. [ ] Confirm feature flags in expected state (if applicable)

**Success Criteria**: All critical journeys working, metrics within SLO, no unexpected errors

**Rollback Trigger**: Critical journey broken, SLO breached, data corruption detected

---

### Step 4: Monitoring Period

**Duration**: [Time to monitor before declaring success - e.g., 2 hours, 24 hours]

**Metrics to Watch**:
- [ ] Request rate (expect: [baseline ± X%])
- [ ] Error rate (expect: < [threshold]%)
- [ ] p95 latency (expect: < [threshold]ms)
- [ ] Database query performance (expect: < [threshold]ms)
- [ ] Error budget consumption (expect: < [X]% of monthly budget)

**Actions**:
- [ ] Monitor dashboards every 15 minutes for first hour
- [ ] Monitor dashboards every hour for next [duration]
- [ ] Review logs for new error patterns
- [ ] Check support channels for user-reported issues

**Success Criteria**: All metrics within expected range, no user-reported critical issues

**Rollback Trigger**: [Specific conditions that require rollback during monitoring]

---

### Step 5: Deployment Completion

**Actions**:
1. [ ] Update status page (deployment complete)
2. [ ] Notify stakeholders (deployment successful)
3. [ ] Document any issues encountered and resolutions
4. [ ] Update runbooks (RUN-XXX) if procedures changed
5. [ ] Schedule post-deployment review (if high-risk deployment)
6. [ ] Clean up old resources (e.g., decommission blue environment after 24 hours)

**Success Criteria**: Deployment marked complete, stakeholders notified

---

## Rollback Plan

### Rollback Decision Tree

**Trigger Conditions** (any of these requires rollback consideration):
- [ ] Error rate > [threshold]% for > [duration] minutes
- [ ] p95 latency > [threshold]ms for > [duration] minutes
- [ ] Critical user journey (UJ-XXX) broken
- [ ] Data corruption detected
- [ ] SLO budget consumption > [X]% in [duration] hours
- [ ] Security vulnerability discovered in deployed code

**Decision Process**:
1. **Assess Impact**: How many users affected? Is it getting worse?
2. **Attempted Fix**: Can we fix forward quickly? (< 15 minutes)
3. **Rollback Decision**: If fix > 15 minutes or impact severe, rollback

**Decision Makers**: [On-call lead, Engineering Manager, CTO (for critical deployments)]

---

### Rollback Procedure

**Strategy-Specific Rollback**:

**Blue-Green**:
1. [ ] Switch traffic back to blue environment (< 1 minute)
2. [ ] Verify blue environment healthy
3. [ ] Monitor for 15 minutes
4. [ ] Investigate issue in green environment (offline)

**Canary**:
1. [ ] Stop deployment (prevent further rollout)
2. [ ] Route traffic away from canary nodes
3. [ ] Revert canary nodes to previous version
4. [ ] Monitor for 15 minutes

**Rolling**:
1. [ ] Stop deployment (prevent rollout to remaining nodes)
2. [ ] Revert deployed nodes to previous version (one AZ at a time)
3. [ ] Monitor each AZ after revert
4. [ ] Verify all nodes on previous version

**Feature Flag**:
1. [ ] Turn feature flag OFF (< 1 minute)
2. [ ] Verify feature disabled for all users
3. [ ] Monitor for 15 minutes
4. [ ] Investigate issue with feature code (offline)

**Database Migration Rollback**:
1. [ ] Run reverse migration script (TEST IN STAGING FIRST)
2. [ ] Verify data integrity
3. [ ] Revert application code to previous version
4. [ ] Monitor for data consistency issues

**Estimated Rollback Time**: [e.g., 5 minutes (feature flag), 15 minutes (code rollback), 1 hour (database rollback)]

---

### Post-Rollback Actions

1. [ ] Update status page (deployment rolled back)
2. [ ] Notify stakeholders
3. [ ] Preserve logs and metrics from failed deployment
4. [ ] Create incident report (if SLO impacted)
5. [ ] Schedule post-mortem (if high-impact)
6. [ ] Fix identified issues in development environment
7. [ ] Re-test before next deployment attempt

---

## Go/No-Go Criteria

**Go Criteria** (all must be true):
- [ ] All pre-deployment checklist items complete
- [ ] All tests passing in staging
- [ ] On-call team available and ready
- [ ] Rollback plan tested and ready
- [ ] No active incidents (P0 or P1)
- [ ] Not during high-traffic period (unless scheduled maintenance)
- [ ] Stakeholders notified and aligned

**No-Go Criteria** (any of these blocks deployment):
- [ ] Tests failing in staging
- [ ] Active P0/P1 incident
- [ ] On-call team unavailable
- [ ] Rollback plan not tested
- [ ] Database backup failed
- [ ] Dependency service unstable
- [ ] High-traffic period (Black Friday, product launch, etc.) without explicit approval

---

## Monitoring & Alerting

**Dashboards to Monitor**:
- [MON-XXX]: [Dashboard name - e.g., API Health Dashboard]
- [MON-YYY]: [Dashboard name - e.g., Database Performance Dashboard]

**Alerts to Watch**:
- [MON-XXX]: [Alert name - e.g., High Error Rate Alert]
- [MON-YYY]: [Alert name - e.g., Latency SLO Breach Alert]

**Silenced Alerts** (during deployment window):
- [Alert name]: [Reason for silencing - e.g., Expected brief downtime during blue-green switch]

**Runbooks**:
- [RUN-XXX]: [Runbook for common deployment issues]
- [RUN-YYY]: [Runbook for rollback procedure]

---

## Communication Plan

**Before Deployment**:
- [ ] [Target audience]: [Message - e.g., "Scheduled maintenance 2026-01-15 02:00-04:00 UTC, expect brief downtime"]
- [ ] Internal teams: Engineering, Product, Support, Sales (if customer impact)
- [ ] External: Customers via email/status page (if downtime expected)

**During Deployment**:
- [ ] Post updates in #deployments Slack channel every 15 minutes
- [ ] Update status page if customer-facing impact

**After Deployment**:
- [ ] [Target audience]: [Message - e.g., "Deployment complete, new features now available"]
- [ ] Internal teams: Deployment summary (duration, issues, outcome)
- [ ] External: Customers via email/status page (if notified beforehand)

**Rollback Communication**:
- [ ] Immediate update in #incidents channel
- [ ] Update status page (if customer-facing)
- [ ] Notify stakeholders of rollback and next steps

---

## Linked IDs

**Related Specifications**:
- [FEA-XXX]: [Features being deployed]
- [API-XXX]: [API contracts affected]
- [DBT-XXX]: [Database migrations included]
- [ARC-XXX]: [Architecture changes]

**Related Operations**:
- [RUN-XXX]: [Runbooks for deployment and rollback]
- [MON-XXX]: [Monitoring and alerting]
- [TEST-XXX]: [Tests validating this deployment]

**Related Business**:
- [KPI-XXX]: [KPIs expected to be impacted by this deployment]
- [UJ-XXX]: [User journeys affected]

---

## Post-Deployment Review

**Scheduled**: [Date/time for review - e.g., 2026-01-16 10:00 AM]

**Attendees**: [Deployment owner, on-call engineers, product manager (if high-impact)]

**Agenda**:
1. Review deployment timeline (planned vs actual)
2. Discuss any issues encountered
3. Review metrics (SLO impact, user feedback)
4. Identify improvements for future deployments
5. Update runbooks/procedures if needed

**Outcomes**:
- [ ] Lessons learned documented
- [ ] Action items assigned (if improvements identified)
- [ ] Runbooks updated (if procedures changed)

---

## Version History

| Version | Date | Change | Updated By |
|---------|------|--------|------------|
| 1.0 | YYYY-MM-DD | Initial deployment plan | [Name] |
| | | | |

---

## Notes

[Any additional context, dependencies, or risks worth documenting]

---

*Template: Use this to create DEP-XXX entries in `SoT.DEPLOYMENT.md`*
writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deploymentStars: 322kUpdated Claude CodeCodexVerifiedInstall
xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy.Stars: 119kUpdated Claude CodeCodexVerifiedInstall
childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release.Stars: 118kUpdated Claude CodeCodexVerifiedInstall