The AI Context Layer
Source-grounded pack with Inventory Agent Context, Define a Business Model for Agents, and Design an Agent Context Contract, plus 1 more.
Use this pack when
Choose this pack when the work touches The AI Context Layer 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.
Inventory Agent Context
Use this workflow when inventorying the exact structured facts, policies, workflow history, evidence, memory, and quality signals an agent task needs, including source authority, freshness, sensitivity, audience, and action relevance.
Output Return an agentcontextinventory with task, contextitems, class, requiredfor, source, authority, owner, freshness, provenance, quality, sensitivity, audiences, memoryclass, conflicts, gaps, and behavior.
Define a Business Model for Agents
Use this workflow when defining the compact business model an agent needs for a workflow: canonical objects, identities, relationships, states, owners, rules, decision rights, allowed claims, actions, and escalation boundaries.
Output Return an agentbusinessmodel with task, objects, identifiers, relationships, states, authoritativesystems, owners, operatingrules, decisionrights, audiencerules, actionprerequisites, stopconditions, escalations, and version.
Design an Agent Context Contract
Use this workflow when turning an agent context inventory into reusable infrastructure with source priority, freshness, identity and audience filtering, memory expiry, quality signals, action prerequisites, escalation behavior, and context-use audit events.
Output Return a contextcontract with requestschema, packetschema, sourcepriority, freshnessrules, identityfilter, audiencefilter, inferenceboundaries, memorypolicy, qualitybehavior, actionrequirements, escalations, auditevents, and owner.
Review Agent Context Risk
Use this workflow when auditing an agent workflow for wrong-object, stale-source, permission, inference, memory, data-quality, action-prerequisite, escalation, or context-audit failures before launch or after an incident.
Output Return a contextriskreview with task, replay, objectrisks, sourcerisks, permissionrisks, inferencerisks, memoryrisks, qualityrisks, actiongaps, escalationgaps, auditability, severity, and fixes.
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 Context Layer series.
10 source essays · 4 usable skills · reviewed Jul 2026