replica-recon
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill requires the agent to ingest content from external websites, including help centers and API documentation, creating a surface for indirect prompt injection. \n
- Ingestion points: Step 2 lists various external URLs for the agent to visit. \n
- Boundary markers: There are no explicit instructions to use delimiters or ignore embedded commands in the fetched data. \n
- Capability inventory: The agent writes results to files in the project's local directory. \n
- Sanitization: No sanitization of the external content is requested.\n- [SAFE]: The skill contains explicit rules prohibiting scraping, unauthorized logins, and decompilation. It emphasizes respect for terms of service and focuses on functionality mapping rather than asset theft.
Audit Metadata