Skills
skills/ascend-ai-coding/awesome-ascend-skills/ai-for-science-tf-to-pytorch/Gen Agent Trust Hub

ai-for-science-tf-to-pytorch

Pass

Audited by Gen Agent Trust Hub on Mar 28, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill contains no instructions aimed at bypassing safety filters or overriding agent behavior. The content is strictly technical and focused on model migration.
  • [DATA_EXFILTRATION]: No network operations, hardcoded credentials, or access to sensitive file paths (such as .ssh or .aws) were found. The skill operates locally on model weights and validation arrays.
  • [REMOTE_CODE_EXECUTION]: There are no patterns involving the download and execution of remote scripts or unverified packages.
  • [OBFUSCATION]: All instructions and code are provided in clear text without any encoding, hidden characters, or homoglyph substitutions.
  • [COMMAND_EXECUTION]: The skill includes a Python script scripts/compare_arrays.py which uses standard libraries (numpy) for numerical comparison. It does not perform any dangerous system calls or arbitrary command execution.
  • [DYNAMIC_EXECUTION]: The Python script uses np.load to read validation data. While np.load can be a deserialization vector if allow_pickle is enabled, the script uses the default settings (which default to allow_pickle=False in modern NumPy), and it targets local data files provided by the user for comparison purposes.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 28, 2026, 02:16 AM
Security Audit — agent-trust-hub — ai-for-science-tf-to-pytorch