msa-structure-prediction-pipeline
Pass
Audited by Gen Agent Trust Hub on Sep 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill communicates with official NVIDIA API endpoints (
health.api.nvidia.com) for protein search and structure prediction, which is appropriate for a skill authored by NVIDIA. - [SAFE]: Sensitive credentials like the
NGC_API_KEYare handled securely via environment variables usingos.getenv(), avoiding hardcoded secrets. - [INDIRECT_PROMPT_INJECTION]: The skill processes data returned from NVIDIA NIM APIs to generate structure predictions and save result files. This represents a standard data processing workflow.
- Ingestion points: API response JSON from
health.api.nvidia.cominSKILL.md. - Boundary markers: None.
- Capability inventory: Writes protein sequence alignments and structure files to the local file system using
openandwritetools inSKILL.md. - Sanitization: Filenames are constructed using the structure format provided by the API; no explicit sanitization is shown, but the source is a trusted vendor.
Audit Metadata