The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-supplied marketing assets, such as ad copy, strategy documents, and landing pages, for the purpose of critique and audit.
Ingestion points: User prompts, pasted text, strategy files, and screenshots are ingested to provide context for the simulated advisors (SKILL.md, Section: "Before Starting").
Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from following malicious commands that might be embedded within the content being audited.
Capability inventory: The skill has the ability to read and write project-local markdown files in the .agents/ directory, execute web search tools for research, and generate HTML artifacts (SKILL.md, Sections: "Before Starting", "Live Research Pass", "Action Matrix").
Sanitization: The instructions do not specify any sanitization or filtering of external content before it is interpolated into the agent's reasoning process.
[DYNAMIC_EXECUTION]: In Boardroom mode, the skill generates a dynamic "Action Matrix" as a standalone HTML artifact.
This artifact includes interactive elements such as filter controls and owner badges based on the agent's analysis (SKILL.md, Section: "Action Matrix").
While the skill specifies that the artifact should be self-contained and avoid external calls, the generation of client-side code from user-influenced data represents a standard dynamic execution surface.

marketing-council