ginkgo-cloud-lab
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines workflows that involve processing untrusted user data, creating a potential surface for indirect instructions. \n
- Ingestion points: Users are expected to upload DNA sequences (FASTA) and data templates (CSV, XLSX, PDF, ZIP) as described in SKILL.md and multiple reference files for protocol submission. \n
- Boundary markers: The skill includes a dedicated 'Agent operating procedure' that mandates human approval before protocol execution, asks agents to 'Ask rather than guess any value', and requires verification of all parameters. \n
- Capability inventory: The skill provides detailed instructions for interacting with Ginkgo Cloud Lab's autonomous lab infrastructure, including protocol configuration, ordering, and management. \n
- Sanitization: The instructions do not specify automated data sanitization or input validation for the sequence files, relying instead on manual validation and procedural human-in-the-loop safeguards.
Audit Metadata