security-threat-model
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill operates by ingesting and analyzing content from source code repositories, which are untrusted external data sources. Maliciously crafted content in a repository could attempt to influence the agent's analysis or output.
- Ingestion points: The skill reads all files within a user-provided repository or path to perform its security analysis.
- Boundary markers: The system prompt and user template use structured headers and fenced blocks to separate task instructions from the analyzed code, reducing the likelihood of instruction confusion.
- Capability inventory: The skill is authorized to read repository files for analysis and write the final threat model report to the filesystem.
- Sanitization: The skill includes explicit instructions to redact secrets (API keys, passwords, tokens) and requires evidence anchors for all findings, which serves as a control against deceptive data.
Audit Metadata