Simulated containment actions only. No production systems are modified.
INC-2041
Finance endpoint compromise
Switch scenario · resets workflow
Critical
Agentic layer · advisory

Agentic Analysis

Eight bounded agents read the modelled incident and produce structured, separated findings — facts, assumptions and unknowns — with recommendations, warnings, confidence and an explicit human-review flag. They propose; people decide.

This layer is advisory and human-approved. Agents cannot execute actions, cannot produce offensive or exploit steps, and always separate facts from assumptions and unknowns. Every containment proposal requires recorded human approval before it can even be simulated.

Agent 1
Triage
Agent 2
Blast Radius
Agent 3
Evidence Preservation
Agent 4
Containment Strategy
Agent 5
Business Impact
Agent 6
Decision Authority
Agent 7
Briefing
Agent 8
Execution Guardrail