infra

name

infra

description

Use when deploying, changing cluster or cloud state, debugging a running service or a failed activation, or reaching for kubectl, terraform, aws, gh, deploy-rs, nixos-rebuild or just switch/deploy. Carries the infrastructure-as-code-first rule and when asking permission is required before touching live systems, and routes to per-tool guidance for Nix deployment, Kubernetes, Terraform, AWS and the GitHub CLI.

Infrastructure

The rule

Infrastructure as code first. Propose the change in the IaC repo and let CI/CD apply it. Mutating live infrastructure directly, such as cloud consoles, ad-hoc kubectl edit, or aws writes, is sometimes acceptable for personal infra, but always ask the user for permission first. Never mutate live infra unprompted.

Reading live state is always fine and is usually the right first step: check what is actually deployed before proposing a change to what should be.

Working rules

• Idempotence. Check current state before mutating. Every operation should be safe to re-run.
• Retry transient failures with backoff before giving up. When a subtask is genuinely unrecoverable, deliver the rest and report the gap explicitly.
• Timestamps in reports and logs are ISO8601 UTC.
• Prefer machine-and-human-readable output (--json, -o json, markdown tables) over free-form dumps.
• Resolve tools through the devshell or nix-shell rather than assuming they are on PATH.

Routing

Working with	Read
This flake: build, switch, deploy-rs, remote hosts	nix.md
Clusters, pods, workload debugging	kubernetes.md
Providers, modules, plan/apply	terraform.md
Accounts, credentials, cloud resources	aws.md
Issues, PRs, releases, repo and code search	github.md