marketing
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from untrusted external sources, which could be leveraged to introduce malicious instructions into the agent's context.
- Ingestion points: The skill retrieves information from web searches (COMPETITIVE mode), processes content from external URLs and local files (BRAND mode), and parses user-supplied CSV files or pasted metrics (PERFORMANCE mode).
- Boundary markers: The instructions do not define specific structural delimiters or instruction-isolation techniques for the ingested content, although the inclusion of
references/llm-marketing-failure-modes.mdprovides strong logical heuristics for content validation. - Capability inventory: The skill possesses capabilities to perform web searches and read files across its various operating modes.
- Sanitization: There is no evidence of programmatic sanitization, filtering, or escaping of the ingested content before it is interpolated into the model's active context.
Audit Metadata