sbom-analysis
Pass
Audited by Gen Agent Trust Hub on Jul 2, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes external SBOM and VEX data which can contain untrusted content.
- Ingestion points: The skill ingests CycloneDX, SPDX, and VEX (CSAF) documents for analysis.
- Boundary markers: The skill includes a dedicated 'Prompt Injection Safety Notice' that explicitly warns the agent to ignore instructions embedded in component metadata or descriptions.
- Capability inventory: Analysis is performed using limited tools (Read, Grep, Glob) for report generation; no executable code or network operations are involved.
- Sanitization: The skill employs instructional safeguards to disregard embedded commands rather than technical sanitization of the input documents.
Audit Metadata