The Agent Skills Directory

[DATA_EXPOSURE]: The skill reads from .agents/product-marketing.md and win-loss-analysis files to gather product context and competitive intelligence. This is a primary function of the skill and occurs within the local environment to generate structured sales responses.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted or third-party data from win-loss notes and product marketing files. While this represents a potential injection surface, the skill implements logic within battlecard.js to validate the structure and tone of the resulting content, effectively sanitizing the output against common sales-specific mistakes (e.g., disparaging language or unbacked claims).
Ingestion points: .agents/product-marketing.md, win-loss-analysis output, and user-provided call notes.
Boundary markers: None explicitly defined for the prompt interpolation, but the Node.js tool enforces structural schema on the data.
Capability inventory: Execution of a local Node.js script and local file system reads.
Sanitization: The battlecard.js engine uses regular expression checks (DISPARAGE, ABSOLUTE) and length validation (MIN_ANSWER_CHARS) to filter and flag low-quality or potentially harmful output before presentation.
[EXTERNAL_DOWNLOADS]: The skill references documentation and tools from aajconsult.com and cites a public repository by Corey Haines (coreyhaines31/marketingskills) for structural inspiration. These references are used for informational purposes and do not trigger automatic downloads of executable code.

objection-handling