The Agent Skills Directory

[PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted third-party data such as customer transcripts, survey responses, and community discussions from platforms like Reddit or Hacker News. This introduces an indirect prompt injection surface where instructions hidden within analyzed materials could attempt to influence the agent's behavior.
Ingestion points: Processes external data in Step 1 and Step 2 from multiple source types including community forums and public review sites.
Boundary markers: The instructions do not define specific delimiters or isolation techniques to prevent the agent from following instructions embedded in the source data.
Capability inventory: The skill uses get_marketing_context and list_marketing_references to access internal docs, and supports web_search and web_fetch tools.
Sanitization: No sanitization or filtering of the research data is performed prior to analysis.

customer-research