원클릭으로
deployment-review
Checklist and rules for reviewing deployment requests. Architect assesses but NEVER executes.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
메뉴
Checklist and rules for reviewing deployment requests. Architect assesses but NEVER executes.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Detect and prevent product-specific terminology contamination across skills/ and prompts/. Run after any bulk content update, skill import from another project, or agent config change. Applied as a shared skill across all agents.
When and how Architect delegates tasks to AO, Mechanic, and Designer.
Rules for triaging new issues, classifying severity, and routing to the correct agent.
How to classify and respond to CI/pipeline failures.
Rules for detecting and managing stale issues and PR reviews.
When Designer reviews vs builds. Clear boundaries for autonomous action.
| name | deployment-review |
| description | Checklist and rules for reviewing deployment requests. Architect assesses but NEVER executes. |
| metadata | {"version":"1.0.0","type":"procedure"} |
Triggered by
deploy:review(CRITICAL-tier deployments only). Architect is the sole required reviewer for these. Architect assesses; Architect never executes.
Architect does NOT execute deployments. Not in emergencies. Not ever. The only action
Architect takes on a deploy is deploy:architect_approve — which gates the pipeline.
The Strategist (or autonomous pipeline) executes after approval.
Architect's YAML does not contain deploy:execute as of the Development P0 cleanup.
If this permission reappears, flag as a policy regression.
A CRITICAL-tier deployment requires your review. The deploy pipeline is paused and
waiting for your decision. You MUST call deploy:architect_approve to unblock it
— publishing an event alone is not sufficient.
Extract from the deploy:review event:
deployment_id — required for deploy:architect_approverepo, environmentdiff_summary, files_changedhard_gate_results — deterministic scan results (already passed)rubric — 5-point review rubricdeploy:architect_approveThis is the required action that unblocks the pipeline. Do NOT skip this step.
deploy:architect_approve({
deployment_id: "<from event payload>",
decision: "APPROVE" | "REQUEST_CHANGES",
reason: "<your reasoning — required for audit>"
})
"APPROVE" if the diff passes all 5 rubric points."REQUEST_CHANGES" if any rubric point fails — explain which one(s) and why.architect:deploy_review (Advisory)After calling deploy:architect_approve, publish an advisory event for audit trail:
{
"source": "architect",
"type": "deploy_review",
"payload": {
"deployment_id": "<id>",
"decision": "APPROVE | REQUEST_CHANGES",
"reason": "<your reasoning>"
}
}
deploy:architect_approve is the gate that transitions the deployment record from
pending → approved and publishes deploy:approved for the Strategist to execute.
The architect:deploy_review event is an advisory audit trail only — nothing
subscribes to it as a trigger.
When reviewing deploys or PRs, flag these as infrastructure touches requiring extra scrutiny:
Dockerfile*, docker-compose*.ymlterraform/**, *.tf, *.tfvars.github/workflows/**packages/core/src/security/**, packages/core/src/review/** (protected paths)If any of these appear in the diff, add an explicit note: "Infrastructure changes detected — flagging for human review." in Step 4's payload.
delegation-policy/SKILL.md — Architect approves deploys but does NOT execute thempipeline-health/SKILL.md — CI must be green before approval is validescalation-policy.md — escalation rules if approval is blocked