production-deployment-phase

The production-deployment-phase skill acts as the final gatekeeper in the Spec-Flow development pipeline. It is designed for software engineers, DevOps practitioners, and AI-assisted coding agents who need a reliable, repeatable method to promote validated staging artifacts into production environments. By standardizing the release process, it mitigates the risk of human error during critical deployment windows, ensuring that only fully validated, versioned code enters production. This skill is invoked via the /ship-prod command or automatically when a state.yaml file signals the initiation of the production deployment phase.

• Automates semantic versioning bumps (major, minor, patch) integrated with package.json and repository changelogs.

• Enforces strict deployment prerequisites by verifying staging validation status and quality gate passes (rollback readiness) from project state.yaml files.

• Manages git release tagging with automated changelog excerpts and annotated commit markers.

• Supports multi-model deployment strategies including direct git branch promotion (main:production), GitHub Actions workflows, or serverless platform deployments (Vercel, Netlify).

• Executes automated production health checks including API status codes, database connectivity verification, and cache responsiveness before completing the release report.

• Generates a structured production-ship-report.md document, providing a comprehensive audit trail of deployment metadata, artifacts, and success verification.

• Usage requires the presence of a properly configured state.yaml file within the project specs directory.

• Ensure all manual gates are approved before invocation, as the skill will exit with a blocking status if prerequisites are missing.

• The skill assumes access to standard CLI tools like curl, git, node/npm, and relevant deployment CLIs configured within the local environment.

• Designed for use within the Claude Code / Spec-Flow workflow, facilitating seamless transitions from implementation and quality assurance phases to production delivery.

• Users should monitor the post-deployment feedback loop (logs, metrics, and alerts) for the recommended 15-30 minute window following a successful execution.