content-strategy
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONMETADATA_POISONING
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and analyze untrusted data from various external sources, which could be used to inject malicious instructions into the agent context.\n
- Ingestion points: Processes information from
.agents/product-marketing-context.md, user-provided keyword exports, sales/customer call transcripts, survey responses, and web search results from platforms like Reddit, Quora, and competitor websites (all referenced in SKILL.md).\n - Boundary markers: The instructions do not define delimiters or explicit warnings to isolate untrusted external content from the skill's primary instructions.\n
- Capability inventory: The skill utilizes web search capabilities and file reading (SKILL.md) to gather strategy insights.\n
- Sanitization: No methods for sanitizing, escaping, or validating the external data are mentioned before the information is processed by the agent.\n- [METADATA_POISONING]: The skill's YAML frontmatter in SKILL.md contains a 'risk: critical' field and a 'date_added' set to 2026-03-21. These fields are unusual for a marketing tool and could be considered misleading metadata in a security context.
Audit Metadata