content-refresh
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill fetches external content from target URLs and competitor search results, which introduces a surface for indirect prompt injection where malicious instructions in third-party content could influence the agent's behavior. Ingestion points: Fetch operations for the target page and top 5 results described in Process step 3 of SKILL.md. Boundary markers: Absent. No instructions are provided to the agent to ignore or delimit instructions found in the fetched content. Capability inventory: The agent executes local scripts (scripts/dfs_client.py, scripts/report_publish.py) and writes generated reports to the filesystem. Sanitization: No sanitization of the fetched web content is described before it is used for mapping and rewriting.
- [COMMAND_EXECUTION]: The skill relies on shell commands to run local Python scripts (dfs_client.py, google_client.py, and report_publish.py) using user-supplied parameters such as domains and keywords. Evidence: Python script calls in the 'Data pulls' and 'Client deliverables' sections of SKILL.md.
Audit Metadata