poc-weaponization
Pass
Audited by Gen Agent Trust Hub on Nov 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted third-party artifacts, which serves as a potential vector for malicious instructions embedded within the analyzed source code or metadata.
- Ingestion points: The agent is instructed to read external source code, manifests, lockfiles, and build configurations.
- Boundary markers: The instructions explicitly treat input as 'untrusted code' and require a structured checklist to detect suspicious patterns.
- Capability inventory: The workflow involves file system access for static review and potentially network access for artifact retrieval, though it prohibits unreviewed execution.
- Sanitization: No technical sanitization of analyzed text is specified for the agent's internal processing.
- [EXTERNAL_DOWNLOADS]: The skill references external documentation and a third-party GitHub repository to support artifact verification and replay testing.
- References official artifact verification guidelines from slsa.dev and NIST.
- References a specific replay contract documentation from the originsec/pocsmith repository on GitHub.
Audit Metadata