tooluniverse-structural-variant-analysis
Pass
Audited by Gen Agent Trust Hub on Mar 15, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests untrusted genomic data and clinical phenotypes to perform analysis and query external databases.
- Ingestion points: Processes user-supplied genomic coordinates, gene symbols, and patient clinical features within the analysis phases defined in SKILL.md.
- Boundary markers: The workflow lacks explicit delimiters or instructions for the agent to disregard instructions potentially embedded within clinical data or retrieved search results.
- Capability inventory: The skill utilizes extensive tool sets for searching scientific databases (OMIM, ClinVar, PubMed, etc.) and generates file-based reports (SV_analysis_report.md).
- Sanitization: No explicit sanitization or strict schema validation for input genomic parameters is mandated before tool execution.
- [SAFE]: No evidence of malicious direct prompt injection, credential theft, or unauthorized remote code execution was detected. The skill's logic is consistent with established medical guidelines (ACMG/ClinGen) and interacts solely with well-known, trusted scientific repositories.
Audit Metadata