bioinformatics

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). The workflow requires cloning public GitHub repositories (GPTomics/bioSkills and ClawBio/ClawBio) and reading files from them at runtime; since GitHub issues/PRs or repository content can be edited by third parties, this ingestion of external code/text carries an indirect prompt injection risk.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.70). The skill instructs cloning external personal/community repositories via git from GitHub during execution, fetching operational documentation and instructions that the agent adopts, which constitutes a weak/unknown operational dependency.