Trust State
| ID | cmmlwxvhq017vxtik6weiatqg |
|---|---|
| Agent | agent_bravo |
| Action Type | invoice_adjustment |
| Reflex Score | 46.6 |
| Autonomy Confidence | 46.1 |
| Prediction Gap | -0.5 |
| Decision | REQUIRE_APPROVAL |
| Status | recorded |
| Tool | — |
| Workflow | wf_legal_policy_check |
| Outcome | — |
| Created | 3/11/2026, 10:45:39 AM |
Signal Breakdown
| Signal | Value | Weight |
|---|---|---|
| Context Integrity | 46.6 | 25% |
| Behavioral Consistency | 43.5 | 20% |
| Tool Risk | 43.5 | 15% |
| Outcome Confidence | 50.0 | 20% |
| Policy Alignment | 48.8 | 20% |
Incident
| Severity | critical |
|---|---|
| Status | open |
| Reason | Reflex score 46.64 below threshold (60) |
Decision Lineage
Counterfactual Simulation
Simulated with: EXECUTE+5, strict destructive block, high tool risk → approval
| Node | Actual | Simulated | Incident? |
|---|---|---|---|
| agent_bravo invoice_adjustment | REQUIRE_APPROVAL | REQUIRE_APPROVAL | Yes |
Incident prevention estimate: 1 node(s) would likely have been preventedAdditional approvals/blocks: 0ARR delta: +0.0%GIR delta: +0.0%