agent-deployment-pipeline

CI/CD pipelines for agent deployments automate building, testing, and releasing code through GitHub Actions workflows. This Skill implements a multi-stage pipeline with push triggers, automated testing, evaluation gates, and progressive canary rollouts to reduce manual work and deployment risk.

Evaluation gates are automated safety checkpoints that run tests and validations before deployment. This Skill uses MLflow evaluation to gate decisions, ensuring agents meet safety and performance criteria before staging or production release.

GitOps with ArgoCD enables declarative, version-controlled deployments where infrastructure state lives in Git. This Skill integrates ArgoCD for progressive rollouts and canary deployments, letting you track and roll back changes through Git commits instead of manual commands.

Yes. This Skill automates the entire pipeline from push to production: container builds, testing, safety evaluation, staging deployment, canary rollout, and production release. Evaluation gates ensure only validated agents advance automatically.

Pushing changes to the agents/* folder triggers the GitHub Actions workflow automatically. Monitor pipeline status in GitHub Actions dashboards and view deployment progress in ArgoCD for real-time visibility across build, test, eval, and rollout stages.

This Skill targets GitHub Actions and ArgoCD environments. If you use these platforms, the Skill provides ready-to-use pipeline templates and GitOps workflows; if not, you'll need to adapt the patterns to your CI/CD platform and container registry.