Traces
The Traces tab shows correlated event sequences. Follow a request through the system, trace policy resolution steps, track prompt chains, and view attestation lifecycles.
The Traces tab has five sub-tabs, each showing a different correlation view:
| Sub-tab | Description | Use Case |
|---|---|---|
| Request Flow | End-to-end request journey | Debug failures, understand request path |
| Policy Trail | Policy resolution sequence | Debug policy decisions, understand inheritance |
| Prompt Chains | Prompt creation and derivation | Track prompt lineage, debug prompt issues |
| Agent Lifecycle | Agent registration/unregistration | Monitor agent health, debug connection issues |
| Attestation Chains | Attestation lifecycle events | Track approval workflow, audit access |
Request Flow
Request Flow shows all events associated with a single request, from receipt to completion.
Left Panel: Request List
- • Grouped by request (agent → operation)
- • Shows request ID and timestamp
- • Event count per request
- • Status badge: OK, No exec, Deny
Right Panel: Event Timeline
- • Chronological event sequence
- • Event type with icon
- • Source → Target details
- • Result status
Typical Request Flow
1. request_received → Agent receives incoming request
2. request → Request details logged
3. request_routed → Request routed to target
4. prompt_received → Prompt context received
5. policy_resolve_start → Policy lookup begins
6. policy_fetch (×N) → Each policy in chain fetched
7. policy_resolve_complete → Final policy computed
8. policy_decision → Allow/deny decision made
Policy Trail
Policy Trail shows the policy resolution sequence for each decision. Useful for understanding how inheritance works and debugging unexpected denials.
Understanding Policy Resolution
- •
policy_resolve_start— Resolution begins with principal context - •
policy_fetch— Each policy in chain fetched (shows found/not found) - •
policy_resolve_complete— Final effective policy computed
Prompt Chains
Prompt Chains tracks prompt creation, derivation, and receipt across agents. Useful for debugging prompt issues and understanding prompt lineage.
Event Types
- •
prompt_created— New authenticated prompt generated - •
prompt_derived— Prompt derived from parent - •
prompt_received— Prompt received by agent
Agent Lifecycle
Agent Lifecycle shows registration and unregistration events for each agent. Useful for monitoring agent health and debugging connection issues.
Event Types
- •
agent_registered— Agent connected and registered - •
agent_unregistered— Agent disconnected or deregistered
Attestation Chains
Attestation Chains shows the full lifecycle of each attestation, from creation through approval/denial to access and consumption.
Left Panel: Attestation List
- • Attestation ID
- • Agent → Capability mapping
- • Timestamp and event count
- • Status: approved, denied, expired, disabled, consumed
Right Panel: Event Timeline
- •
created— Request initiated - •
approved/denied— Admin decision - •
accessed— Grant used (may repeat) - •
consumed/expired/disabled— Terminal state
Attestation Lifecycle
created → approved/denied → accessed (×N) → consumed/expired/disabled