Trace authority from sponsor to runtime action.
Build a live identity graph of agents, models, tools, workloads and compute, then resolve ownership, delegation, reachable actions and current status before sensitive requests proceed.
See the full operating context, not only the principal.
Discover
Ingest agents, accounts, roles, tokens, models, workloads, tools, clusters and enterprise metadata.
Resolve
Map ownership, sponsorship, delegation, model dependencies, tool calls, data access and execution paths.
Monitor
Check CPID status, credential proof, workload, model, region, destination and action context at sensitive points.
React
Trigger review, suspend the passport, revoke credentials or contain workloads after material change or anomalous activity.
Identity for AI cannot stop at the model.
Runtime trust depends on the workload image, orchestrator, region, network, keys, compute and attestation context as well as the agent and model.
Who is acting and under whose mandate?
Which model is actually invoked?
Where is the authority executing?
Does the execution environment match policy?
Is the requested action inside the approved boundary?