The Agent Skills Directory

Indirect Prompt Injection Surface: The skill is designed to ingest and process meeting transcripts and external files, which could potentially contain untrusted instructions. * Ingestion points: Processes data from Meeting Transcripts and Knowledge & Files (SKILL.md). * Boundary markers: The skill requires adherence to Cross-Skill Best Practices and defines a strict structured output format for the readout. * Capability inventory: Uses tools to search and fetch transcript summaries and full context for validation of trends. * Sanitization: Instructions limit quotes to 25 words per source and prohibit the invention of calls or outcomes to maintain factual accuracy.
Reference to External URLs: The skill generates links to call recordings as evidence. * Evidence: Cites calls using compact inline Markdown links (SKILL.md). * Context: These links are intended to provide transparency and allow users to verify the analysis, which is a standard part of the skill's intended functionality.

review-rep-call-trends