alterlab-borzoi

Pass

Audited by Gen Agent Trust Hub on Sep 25, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill downloads model weights for the Borzoi sequence-to-function model from upstream repositories (referenced as calico/borzoi). It also references other tools in the AlterLab suite such as alterlab-remote-compute, alterlab-gnomad, and alterlab-clinvar for supplementary data and compute tasks.
  • [COMMAND_EXECUTION]: The skill requires access to Bash(python:*) and Bash(uv:*) to execute the Borzoi model and handle genomic data processing. It specifically uses uv run for managing Python execution environments.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes DNA sequences (FASTA) and genomic coordinates provided by the user or external files. While this constitutes an external data ingestion surface, the risk is minimal as the data is treated as biological sequence input for a deep learning model rather than natural language instructions.
  • Ingestion points: DNA sequences, genome coordinates, and reference FASTA files processed in SKILL.md and references/borzoi_usage.md.
  • Boundary markers: None explicitly defined in the prompt templates.
  • Capability inventory: File system access (Read/Write/Edit) and shell execution (Bash) for running model predictions and processing outputs.
  • Sanitization: Standard handling of genomic data formats (FASTA) is expected by the underlying model.
  • [DYNAMIC_EXECUTION]: The skill involves running complex deep-learning models (Borzoi/Enformer) which ingest long genomic contexts. This execution is managed through standard Python environments and recommended GPU acceleration.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 25, 2026, 02:35 PM