audience-cohort-overlap-checker
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted JSON or CSV data which represents a potential attack surface if the data contains malicious instructions intended to influence the agent's behavior during the reporting phase.
- Ingestion points: Data is ingested in
run.pyusingjson.loads(input_path.read_text())from a user-provided file path. - Boundary markers: There are no explicit delimiters or instructions to the LLM to ignore embedded commands within the processed data.
- Capability inventory: The
run.pyscript performs file system operations (read/write) using thepathlibmodule. - Sanitization: The input is parsed as structured JSON, but the content of the fields is not sanitized for natural language instructions.
Audit Metadata