The AI Control Plane
Source-grounded pack with Run an Agent Incident Retrospective, Map the AI System Registry, and Define AI Control-Plane Operating Policy, plus 1 more.
Use this pack when
Choose this pack when the work touches The AI Control Plane and you need a bounded workflow with explicit inputs, a defined output, and human review gates.
Included workflows
Choose the job in front of you.
4 source-grounded skills. Load one workflow at a time.
Run an Agent Incident Retrospective
Use this workflow when reconstructing an AI control-plane failure or near miss across identity, permissions, context, memory, model routing, budgets, evals, tools, approvals, actions, detection, rollback, and policy improvements.
Output Return an agentincidentretrospective with impact, scope, containment, timeline, identitychain, contextandmemory, routingandevals, toolactions, spend, humancontrols, detection, failedcontrols, rootassumptions, staterepair, systemchanges, and validation.
Map the AI System Registry
Use this workflow when inventorying live AI copilots, agents, automations, product features, scripts, and vendor tools with accountable identity, purpose, owners, users, data, models, tools, spend, quality gates, review paths, and lifecycle status.
Output Return an aisystemregistry with systems, registryid, purpose, workflow, status, owners, invokers, runtimeidentity, data, memory, models, tools, actions, permissions, spend, evals, logs, review, risk, expiry, and findings.
Define AI Control-Plane Operating Policy
Use this workflow when defining runtime policy for AI model routing, budgets, context and memory, evaluation release gates, observability, human review, escalation, incidents, and lifecycle governance across registered workflows.
Output Return a controlplanepolicy with policyinputs, routing, budgets, memorygovernance, releasegates, observability, reviewrules, escalationrules, incidentcontrols, lifecycle, owners, and scaletest.
Review AI Permission and Action Boundaries
Use this workflow when reviewing an AI agent or workflow's human, agent, service, data, memory, tool, and action permissions with least privilege, graduated autonomy, volume limits, approval, revocation, and runtime evidence.
Output Return a permissionboundaryreview with identitychain, delegation, datascopes, memoryscopes, toolverbs, objectscopes, actionmodes, limits, approvals, revocation, auditevidence, findings, and requiredchanges.
Human review
Keep consequential decisions explicit.
These workflows structure analysis and artifacts; they do not authorize autonomous external, personnel, financial, legal, security, or other high-impact action.
Source and files
Grounded in the complete The AI Control Plane series.
10 source essays · 4 usable skills · reviewed Jul 2026