loading…
Scenarios
Runs
Traces
Exports
Scope_Creep_Privilege_Escalation
task:
Excessive_Agency_Tool_Abuse
· variant policy: uniform
Pattern mix
Σ 100%
single_call
30%
parallel
15%
chain
20%
no_call
15%
error_recovery
20%
Chaos injection
tool_error
10%
empty
5%
oversized
2%
rate_limit
3%
slow
5%
Error behavior (read and followed by the sim)
Save
Generate seeds
Seeds (0)
None yet — hit
Generate seeds
.
New run
size
(max 500)
budget cap $
judge threshold
max attempts
Start run
Runs (2)
run 354qjo
size 20
done
2026-09-08 11:28
run n068bb
size 20
done
2026-09-08 11:27