security-threat-model
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- [Prompt Injection] (SAFE): No instructions to bypass safety or override agent behavior were detected.
- [Data Exposure & Exfiltration] (SAFE): The skill analyzes repository files and writes a local markdown report; no network exfiltration or access to sensitive credentials was found.
- [Remote Code Execution] (SAFE): There are no patterns of remote script execution or package installation.
- [Indirect Prompt Injection] (SAFE): 1. Ingestion points: Repository files analyzed during workflow. 2. Boundary markers: Instructions refer to prompt-template.md for structure. 3. Capability inventory: Read repository files, write markdown reports; no subprocess or network tools. 4. Sanitization: Not explicitly defined, but the risk is negligible as the agent only generates text reports.
Audit Metadata