Category: Map
Subcategory: MAP 1.1 Official Requirement
AI systems are inventoried and documented. The organization should maintain a current inventory of AI systems, including their purpose, capabilities, dependencies, data sources, intended use cases, and associated risks. Documentation should be sufficient to enable informed risk management decisions and traceability throughout the AI lifecycle. How Katyar Addresses This Requirement Katyar automatically maintains a real-time, auditable inventory of all AI agents through SDK registration and self-discovery, providing complete visibility into deployed systems. Evaluation Criteria
Katyar considers the control satisfied when:
- At least 1 agent has been formally registered and onboarded in the workspace.
- Total number of registered agents
- Agent metadata: name, framework (LangChain, CrewAI, custom, etc.), tools used, onboarded timestamp
- Last seen/heartbeat timestamp (shows active usage)
- Agent status (connected, disconnected, revoked)
- Associated API keys and session history
-
Automatic Agent Registration
Agents self-register on first connection viakatyar.init()or session creation — no manual entry required. -
Rich Agent Metadata
Captures: agent name, framework, tools list, risk level (if set), creation time, last activity — all queryable. -
Centralized Agent Dashboard
Single view lists all agents with status, last seen, tool usage summary, and compliance shield indicator. -
Onboarding Enforcement
Only onboarded agents (via SDK) contribute to compliance scores and inventory — prevents shadow AI from being invisible. -
Export & Reporting
One-click CSV/JSON export of agent inventory for auditors, including timestamps and metadata. -
Real-time Monitoring
Agent connections/disconnections appear in live event stream; CLI commandkatyar agents listshows current inventory.
- Onboard at least one agent using the SDK:
