From release-manager
Plan a release — scope assessment, readiness gates, deployment strategy, communication plan, and go/no-go decision.
npx claudepluginhub hpsgd/turtlestack --plugin release-managerThis skill is limited to using the following tools:
Plan a release for $ARGUMENTS.
Provides UI/UX resources: 50+ styles, color palettes, font pairings, guidelines, charts for web/mobile across React, Next.js, Vue, Svelte, Tailwind, React Native, Flutter. Aids planning, building, reviewing interfaces.
Fetches up-to-date documentation from Context7 for libraries and frameworks like React, Next.js, Prisma. Use for setup questions, API references, and code examples.
Calculates TAM/SAM/SOM using top-down, bottom-up, and value theory methodologies for market sizing, revenue estimation, and startup validation.
Plan a release for $ARGUMENTS.
The cardinal rule: every release is reversible, communicated, and gated. No release ships without passing all gates, having a rollback plan, and briefing support. Pressure does not override process.
Identify everything included in this release:
List changes since last release — use git log to enumerate:
git log --oneline --no-merges <last-release-tag>..HEAD
Categorise each change:
| Category | Includes |
|---|---|
| Features | New user-facing capabilities |
| Enhancements | Improvements to existing features |
| Bug fixes | Corrections to existing behaviour |
| Infrastructure | CI/CD, deployment, config changes |
| Migrations | Database schema changes, data migrations |
| Security | Auth changes, dependency patches, vulnerability fixes |
Identify dependencies — does any change depend on a migration, config change, feature flag, or external service deployment?
Flag risks — which changes have the widest blast radius? Which touch auth, payments, or data?
Confirm scope with product — the release manager does not decide what ships. Confirm the scope matches what the product owner approved
Evaluate every gate. A skipped gate is a failed gate.
Engineering gates:
Operational gates:
Communication gates:
Rules:
Choose a strategy based on the release content and risk profile:
| Strategy | When | Risk | Rollback speed |
|---|---|---|---|
| Feature flag | New features, uncertain impact | Lowest | Instant (toggle off) |
| Percentage rollout | User-facing changes, want to monitor | Low | Fast (reduce to 0%) |
| Blue/green | Infrastructure changes, zero-downtime required | Low | Fast (switch traffic) |
| Canary | High-risk changes, need real traffic validation | Medium | Moderate (redirect traffic) |
| Big bang | Small changes, internal tools, low risk | Higher | Slow (full redeploy) |
Default to feature flags for user-facing changes. Big bang only for low-risk internal changes.
If multiple strategies apply to different parts of the release, combine them. A release can use feature flags for user-facing features and big bang for internal tooling changes simultaneously.
Define BEFORE deployment what triggers a rollback. These are non-negotiable:
| Signal | Threshold | Action |
|---|---|---|
| Error rate | >2x baseline for 5 minutes | Automatic rollback |
| p95 latency | >3x baseline for 5 minutes | Investigate, rollback if not resolving |
| Support ticket spike | >3x normal rate within 1 hour | Investigate, rollback if product-related |
| Health check failures | Any health endpoint returning non-200 | Immediate rollback |
| Data integrity | Any data corruption signal | Immediate rollback + incident response |
Rules:
Identify who needs to know what and when:
| Audience | What they need | When | Channel |
|---|---|---|---|
| Support team | FAQ, known issues, escalation paths, expected behaviour changes | Before deployment | Briefing + document |
| Engineering | Deploy time, what to monitor, rollback plan, on-call expectations | Before deployment | Team channel |
| GTM / Marketing | Feature availability, timing for announcements, any launch coordination | Before deployment | Coordination meeting |
| Customers | What changed, what they need to do (if anything), where to get help | After deployment verified stable | Release notes / email / in-app |
| Leadership | Release summary, risk assessment, go/no-go decision | Before deployment | Status update |
Rules:
Make the call based on gates and evidence:
GO: All engineering, operational, and communication gates pass. Rollback plan verified. Team available to monitor post-deployment.
NO-GO: Any engineering gate fails. No rollback plan. Support team not briefed. Deploying into a known-bad state (existing incident in progress).
CONDITIONAL GO: Some non-engineering gates have known acceptable risks. Document the risk, get CTO approval, proceed with enhanced monitoring. Engineering gate failures are never conditionally acceptable.
Deliver a release plan in this format:
## Release: [version/name]
### Scope
| Change | Category | Risk | PRs |
|---|---|---|---|
| [description] | Feature/Fix/Infra | Low/Med/High | #123 |
### Readiness
| Gate | Status | Evidence |
|---|---|---|
| Tests pass | PASS/FAIL | [command + exit code] |
| Security review | PASS/FAIL/N/A | [reviewer + date] |
| Migrations tested | PASS/FAIL/N/A | [staging verification] |
| Performance baseline | PASS/FAIL | [benchmark results] |
| Rollback tested | PASS/FAIL | [method + verification] |
| Support briefed | PASS/FAIL | [date + document link] |
| Docs updated | PASS/FAIL/N/A | [what was updated] |
### Strategy: [Feature flag / Percentage rollout / Blue-green / Canary / Big bang]
Reasoning: [why this strategy fits this release]
### Rollback Criteria
| Signal | Current baseline | Threshold | Action | Owner |
|---|---|---|---|---|
| Error rate | [value] | >2x for 5 min | Automatic rollback | [name] |
| p95 latency | [value] | >3x for 5 min | Investigate + rollback | [name] |
### Communication
| Audience | Status | Date |
|---|---|---|
| Support | Briefed/Pending | [date] |
| Engineering | Notified/Pending | [date] |
| Customers | Drafted/Pending | [date] |
### Decision: [GO / NO-GO / CONDITIONAL GO]
Reasoning: [why, citing gate results]
/release-manager:rollback-assessment — before releasing, define the rollback plan. Every release plan should have a corresponding rollback assessment.Track DORA metrics across releases: deployment frequency, lead time, change failure rate, and time to restore. These are the industry-standard measures of release process health.
Use the release checklist template (templates/release-checklist.md) and rollback plan template (templates/rollback-plan.md) for consistent output structure.