nvshmem-collect-performance-data
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Executes NVSHMEM benchmarks and a bundled system collection script (
scripts/collect-performance-environment.sh) to gather diagnostic data. The execution is scoped to the user's current compute allocation and does not require elevated privileges. - [INDIRECT_PROMPT_INJECTION]: The skill processes user input for benchmark selection and parses output from external tools to generate reports. This surface is managed by strict validation and use of pre-defined templates.
- Ingestion points: User benchmark choices and command output from executed performance tests (
SKILL.md). - Boundary markers: Employs a dedicated Markdown template (
assets/performance-report-template.md) to structure and isolate the collected data. - Capability inventory: Shell execution of performance benchmarks and system diagnostic utilities (
SKILL.md,scripts/collect-performance-environment.sh). - Sanitization: Validates benchmark output for numeric consistency, preserves native units, and verifies successful process exit codes.
Audit Metadata