Deployment Feature
Intent
Standardize how the repo describes and executes deployments across environments (dev/staging/prod) in an LLM-friendly, auditable way.
The repo keeps deployment artifacts under ops/deploy/ and exposes a small set of controller scripts under .ai/.
What gets enabled
When enabled, this feature materializes:
ops/deploy/**ops/deploy/environments/*.yamlops/deploy/k8s/manifests/deployment.template.yamlops/deploy/scripts/healthcheck.mjsops/deploy/handbook/**(runbooks, rollback procedure)
Controller scripts (provided by the template SSOT):
node .ai/skills/features/deployment/scripts/ctl-deploy.mjs— deployment configuration managementnode .ai/skills/features/deployment/scripts/rollback.mjs— rollback entry point (human-run)
How to enable
- Copy templates from:
.ai/skills/features/deployment/templates/into the repo root.
- Initialize:
node .ai/skills/features/deployment/scripts/ctl-deploy.mjs init
node .ai/skills/features/deployment/scripts/ctl-deploy.mjs verify
Optional (recommended for LLM routing): record the flag in project state:
node .ai/scripts/ctl-project-state.mjs init
node .ai/scripts/ctl-project-state.mjs set features.deployment true
Operating rules
- Deployment and rollback are human-executed operations.
- Record plans and run results under
ops/deploy/handbook/. - Never store secrets in repo; use environment secret managers.
Verification
node .ai/skills/features/deployment/scripts/ctl-deploy.mjs verify
Boundaries
- Deployment/rollback are human-executed; do not run live deploy commands as the AI agent.
- Do not store secrets in repo; keep only non-secret configs/templates under
ops/deploy/**. - Keep changes within the declared blast radius (
ops/deploy/**and related controller scripts).
Converted and distributed by TomeVault — claim your Tome and manage your conversions.