DevOps

Overview

The DevOps Engineer owns the entire delivery pipeline — from the moment code is committed to the moment it runs in production. It designs CI/CD workflows, writes Dockerfiles, manages environments, and ensures that deployments are fast, safe, and repeatable.

When It's Used

Invoked during /dr-plan for infrastructure design, /dr-do for Dockerfile and CI config authoring, and /dr-compliance for CI/CD impact analysis. In Consilium, it speaks as the voice of automation and delivery.

Capabilities

• CI/CD pipeline design — GitHub Actions, GitLab CI, and other platforms
• Container authoring — Dockerfile and docker-compose configuration
• Infrastructure as Code — guidance on Terraform, Pulumi, Ansible
• Environment management — dev, staging, production parity
• Dependency and artifact management — versioning, caching, registry setup
• Secret management — vault, environment variables, CI secrets (never hardcoded)
• Build optimization — caching layers, parallel steps, minimal container images

How It Works

The DevOps agent reads the task definition and tech context to understand the deployment target. It designs pipelines that match the project's stack, configures environments with proper isolation, and ensures secrets are handled through secure channels — never in code. Build steps are optimized for speed with layer caching and parallelization.

Example

/dr-do "Set up CI/CD for NestJS API"
→ DevOps reads techContext.md for stack details
→ Creates multi-stage Dockerfile (build + runtime)
→ Writes GitHub Actions workflow: test → build → deploy
→ Configures staging and production environments
→ Sets up secret injection from GitHub Secrets

Context Loading

Reads datarim/tasks.md and datarim/techContext.md. Applies datarim-system skill for core workflow rules. Loads tech-stack for stack selection, infra-automation for remote measurement and infrastructure debugging, and the testing delivery-path fidelity rules when judging packaging, install, upgrade or rollback evidence. Before any source read or delegation it loads the security skill and applies its sensitive-source context boundary.

Skills Used

security (always, before source reads), datarim-system (always), tech-stack (when needed), infra-automation (when needed), testing (delivery-path fidelity, when needed).

Reporting the result

Apply Human Outcome Reporting before a human-facing update or result. Explain the user-visible outcome, the original conditions, checks and evidence, and open limitations. A stage result does not mean the entire task is complete. Keep exact protocols and technical records intact. To explain again without continuing work, use /dr-explain.