platform-deploy

name	platform-deploy
description	Platform Developer: Stage 8 deploy task. Adapter-driven deployment — read config.yml for the adapter, follow adapter instructions, write deploy-log.md and stage-08.json.

Platform Deploy Task (Stage 8)

Use this skill when you are the Platform Developer executing the Stage 8 deploy stage. Stage 8 is adapter-driven. Read .devteam/config.yml, discover which adapter the project has selected, and follow that adapter's instructions in .devteam/adapters/<adapter>.md. If no adapter is configured, default to local and follow .devteam/adapters/local.md.

Step 0 — Common preconditions (every adapter)

1. PM sign-off. Read pipeline/gates/stage-07.json. If "pm_signoff": true is absent or false: write "status": "ESCALATE" with reason "PM sign-off missing — cannot deploy" and halt. If "deploy_requested": false, do not ask the Principal whether to deploy. Treat it as an explicit instruction that no external deployment was requested; use the local adapter, run/record local verification, and mark adapter_result.external_deploy: false.
2. Runbook. Confirm pipeline/runbook.md exists and contains at minimum a ## Rollback and ## Health signals section. If missing: write "status": "ESCALATE" with reason "Runbook required for Stage 8".
3. Config. Read .devteam/config.yml. Find deploy.adapter. If missing, use local. Accept one of: local, docker-compose, kubernetes, terraform, cloud-run, gizmos, npm, custom. Unknown adapter: write "status": "FAIL" with reason "Unknown deploy adapter." Do not escalate for a missing adapter; local is the safe default.

Step 1 — Load adapter instructions

Read .devteam/adapters/<adapter>.md and follow the adapter's numbered procedure. Adapters are authoritative for their own deploy story.

Step 2 — Write outputs

Every adapter's procedure ends with writing two artefacts:

1. pipeline/deploy-log.md: human-readable record of the deploy, including a **Runbook**: pipeline/runbook.md §<section> line that points a future on-call engineer at the recovery procedure.
2. pipeline/gates/stage-08.json: gate with the baseline fields required by .devteam/rules/gates-core.md plus:

   {
     "deploy_adapter": "<name>",
     "environment": "<env>",
     "smoke_test_passed": true,
     "runbook_referenced": true,
     "adapter_result": { /* adapter-specific */ }
   }
   

Step 3 — Failure handling

On any step failure: write "status": "FAIL" with the failing output as a blocker, halt. Do NOT auto-rollback. The runbook names the rollback procedure and the orchestrator surfaces it to the user; a human decides whether to roll back immediately or investigate first.

The user can follow the runbook's §Rollback section. Do not execute rollback from the role unless the adapter explicitly declares auto-rollback is safe for it (none of the built-in adapters do).