accelerated-computing-cudf

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [DYNAMIC_EXECUTION]: The skill includes an evaluation script (evals/files/cudf-native-stream-handoff-boundary/code/run_smoke.sh) that performs runtime compilation of a local CUDA C++ source file (threaded_handoff.cu) using the nvcc compiler, followed by execution of the binary to verify GPU memory handoff logic. This is standard behavior for accelerated computing development skills.
  • [INDIRECT_PROMPT_INJECTION]: The skill includes numerous examples for ingesting and processing external data from CSV and Parquet files. This represents a standard data processing surface typical of data science skills, though no specific bypass or injection vulnerabilities were identified.
  • [EXTERNAL_DOWNLOADS]: The skill references official NVIDIA and RAPIDS documentation, API references, and GitHub repositories hosted on docs.rapids.ai and github.com. These are legitimate vendor resources used for providing technical depth.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 09:08 PM
Security Audit — agent-trust-hub — accelerated-computing-cudf