Simulated environment. The company, hosts and people are fictional. Most of the attacker infrastructure and user agents come from public ARTEX reporting. Two values do not: the user agent artex-probe/0.4 (+planner), and 203.0.113.116, a reserved documentation address standing in for one exfiltration destination.
A replay of the lab's real 7AI investigations in the 7AI interface: the same tool runs, requests, responses, timings, conclusions and recommendations, with the company renamed. The clock follows the recorded tool timestamps, and idle stretches play faster. Approval controls are illustrative. In 7AI, each response action runs automatically or waits for approval based on the authority you set.
A replay of the lab hunt in the 7AI interface: the analyst's request, every query 7AI wrote with its intent, request and result, and the final report, with the company renamed. Query timing within each round is approximate; round times and query durations are from the run. Containment approvals are illustrative.
Vet before you block. The FBI notes these addresses were observed historically and that cloud-hosted IPs get reassigned. A match is a lead to investigate, not proof of compromise, and an agent can change its address and user agent in minutes.
| Indicator | Type | Context | Source |
|---|
Indicators are shown defanged. Copy and CSV export give the original values, ready for a query.