viennarna-structure-prediction

Pass

Audited by Gen Agent Trust Hub on Jul 10, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the user to install the viennarna package via conda-forge and bioconda channels, and scientific libraries (matplotlib, numpy, pandas) via pip. These are well-known, established package repositories and channels used extensively in the scientific community.
  • [COMMAND_EXECUTION]: The skill uses subprocess.run to interface with the RNAfold command-line tool. The implementation uses a list-based argument structure and provides the RNA sequence via standard input, which is a secure practice that prevents command injection.
  • [PROMPT_INJECTION]: The instructions are focused strictly on the domain of bioinformatics and RNA analysis. There are no attempts to override agent behavior, bypass safety filters, or extract system prompts.
  • [DATA_EXFILTRATION]: No network operations or attempts to access sensitive files (such as credentials, SSH keys, or environment variables) were found. The skill operates locally on provided RNA sequences.
  • [SAFE]: The skill includes input validation in the prepare_sequence function, ensuring that input strings only contain valid RNA nucleotide characters ('ACGUNX') before processing. This reduces the surface for injection or parsing errors in the underlying C library bindings.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 10, 2026, 03:35 PM
Security Audit — agent-trust-hub — viennarna-structure-prediction