nextflow-development

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

• Third-party content exposure detected (high risk: 0.90). The skill's required workflow (SKILL.md Step 0 and references/geo-sra-acquisition.md) explicitly instructs the agent to fetch study info and download datasets from public GEO/SRA/ENA (NCBI/ENA) and to read that metadata to suggest genomes/pipelines and sample subsets, so untrusted, user-generated third-party content is ingested and can materially influence decisions.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill includes an installation command that would fetch and execute remote code at runtime via "curl -s https://get.nextflow.io | bash" (the get.nextflow.io URL), which is a direct remote-code-execution vector if followed as part of setup.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 1.00). The prompt includes explicit sudo commands and instructions to change group membership, start system services, and install packages (e.g., sudo usermod -aG docker $USER, sudo systemctl start docker, sudo apt install ...), which instruct the agent to perform privileged system modifications that alter the machine state.