Runs an automated quality-gate review across multiple lanes (code, docs, tests, security) before approving a release.

Install

mkdir -p .claude/skills/release-serithemage && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/9534" && unzip -o skill.zip -d .claude/skills/release-serithemage && rm skill.zip

Installs to .claude/skills/release-serithemage

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

Runs a comprehensive release review before tagging and publishing a new version. Reviews code, docs, tests, security, cost, and operations; use platform-native parallelism when available and a sequential fallback otherwise.
223 chars✓ has a “when” trigger
Advanced

Key capabilities

  • Run release review
  • Review code, docs, and tests
  • Perform security and cost audits
  • Publish release

How it works

It executes a six-lane review process (code, docs, tests, security, cost, ops) before tagging and publishing.

Inputs & outputs

You give it
Target version
You get back
Published release

When to use release

  • Pre-release quality gates
  • Verifying release notes
  • Running regression tests
  • Conducting final security audit

About this skill

Release Review & Publish

Target version: $ARGUMENTS

Procedure

Run the six release lanes below, aggregate the evidence, and block release if any lane rejects.

Step 1: Pre-flight Checks

Before lane review, verify basics:

npm run build
npm run lint
npm run test
npm run test:e2e

All must pass before proceeding.

Step 2: Review Lanes

If the platform supports subagents or workers, run independent lanes in parallel. Otherwise, run them sequentially.

Each lane must return APPROVE or REJECT with evidence.

Lane 1: Code Review

Scope: git diff v{previous-tag}...HEAD

Checklist:

  • no debug leftovers, TODOs, or silent catches
  • types are sound and any is justified
  • no backward-incompatible changes without documentation
  • no secrets, credentials, or unsafe import/runtime patterns

Lane 2: Documentation Review

Scope: docs/, README.md, CLAUDE.md, RELEASE_NOTES.md

Checklist:

  • docs remain English-only
  • architecture and test counts match current behavior
  • release notes contain the target version
  • broken links or stale commands are corrected

Lane 3: Test Coverage Review

Scope: all changed tests and their neighboring source files

Checklist:

  • new behavior has regression coverage
  • no skipped tests or obviously flaky patterns
  • edge cases and failure paths are covered

Lane 4: Security Review

Scope: source, CDK, workflows, Dockerfiles

Checklist:

  • no secrets in code or images
  • IAM remains least-privilege
  • Bridge auth, IDOR prevention, and SSM secret resolution stay intact
  • no suspicious changes to release or security boundaries

Lane 5: Cost Review

Scope: infrastructure and runtime configuration

Checklist:

  • no NAT Gateway, ALB, or Interface Endpoint regressions
  • PAY_PER_REQUEST remains in place
  • log retention and idle-cost controls stay sane
  • Lambda-only mode does not accidentally reintroduce fixed Fargate cost

Lane 6: Operations Review

Scope: deployability, rollback, monitoring, and runbooks

Checklist:

  • deployment and rollback paths are documented
  • dashboards, watchdogs, and health probes still make sense
  • make targets and operational commands still match the repo

Step 3: Aggregate Results

Collect a compact table of lane verdicts and findings. Release only if all lanes approve and no critical or high-severity issue remains.

Step 4: Publish

Only after approval:

git add -A
git commit -m "chore: prepare release $ARGUMENTS"
git push origin main
gh release create "$ARGUMENTS" --title "$ARGUMENTS" --notes-file RELEASE_NOTES.md --target main

References

When not to use it

  • When the project is not ready for release
  • When the user wants to bypass release gates

Prerequisites

Release version

Limitations

  • Requires all lanes to pass
  • Rigorous process

How it compares

It enforces a rigorous, multi-lane quality gate for releases, blocking publishing if any lane fails.

Compared to similar skills

release side by side with the closest alternatives in the catalog.

SkillInstallsUpdatedSafetyDifficulty
release (this skill)04moReviewAdvanced
code-review-checklist325dNo flagsBeginner
springboot-verification44moReviewIntermediate
kb-check02moNo flagsIntermediate

Try saying

Example prompts that trigger this skill in your AI assistant.

Search skills

Search the agent skills registry