blender-agent-benchmark
Warn
Audited by Socket on Oct 5, 2026
1 alert found:
AnomalyAnomalyscripts/run_benchmark.ts
LOWAnomalyLOW
scripts/run_benchmark.ts
No clear malicious behavior or intentional data theft is present. This is a benchmark runner that intentionally launches an AI agent and Blender, but its full-access agent configuration and execution of generated Python create substantial execution risk if inputs, tools, or agent output are untrusted.
Confidence: 97%Severity: 67%
Audit Metadata