cpu-gpu-performance
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute standard system monitoring and profiling commands including
uptime,ps,nvidia-smi,perf,nsys, andnvprof. These tools are used to establish performance baselines and instrument workloads. - [INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it ingests and processes output from external system commands and test frameworks.
- Ingestion points: Output from profiling tools (
uptime,ps,nvidia-smi) and build/test tools (pytest,bazel,cargo). - Boundary markers: Not explicitly defined in the prompt templates.
- Capability inventory: The skill possesses the capability to execute local shell commands for performance monitoring and testing.
- Sanitization: No explicit sanitization or validation of the command output is described before the agent logs or interprets the data.
Audit Metadata