Skills
Navigate
skills/omicverse/omicverse-skills/bulk-rna-seq-batch-correction-with-combat/Gen Agent Trust Hub

bulk-rna-seq-batch-correction-with-combat

Warn

Audited by Gen Agent Trust Hub on Aug 6, 2026

Risk Level: MEDIUMREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill instructions and reference code in SKILL.md and reference.md encourage the use of pd.read_pickle() to load data. The Python pickle module is known for unsafe deserialization, which allows for arbitrary code execution if an attacker provides a maliciously crafted pickle file.
  • [PROMPT_INJECTION]: The skill ingests untrusted data from external files and performs file-writing operations, representing a surface for indirect prompt injection.
  • Ingestion points: pd.read_pickle() and pd.read_csv() in SKILL.md and reference.md.
  • Boundary markers: None present to instruct the agent to ignore instructions embedded in the expression matrices.
  • Capability inventory: The skill has file-writing capabilities through to_csv() and write_h5ad() in SKILL.md and reference.md.
  • Sanitization: No sanitization or data validation steps are implemented for the ingested data.
  • [SAFE]: The skill references documentation and examples from omicverse.readthedocs.io, which is an established and well-known documentation service for the relevant bioinformatics tools.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 6, 2026, 10:28 AM