The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill allows for a 'Live Research Pass' that fetches data from the internet via web search or specialized tools like deep-research. This untrusted data could contain malicious instructions designed to subvert the agent's behavior.
Ingestion points: Live research results from native web search, deep-research, watch-video, and last30days tools are ingested into the agent's context (specified in SKILL.md).
Boundary markers: Absent. The instructions do not define specific delimiters or instructions to ignore commands embedded in the retrieved web content.
Capability inventory: The skill is primarily designed for analysis and strategy; however, it has the ability to write custom advisor dossiers to the local .agents/advisors/ directory and can hand off tasks to other functional skills like copywriting or ads (specified in SKILL.md).
Sanitization: No explicit data sanitization or filtering logic is prescribed for the external research findings beyond the requirement to provide citations.

marketing-council