The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted data from external sources, which may contain malicious instructions intended to manipulate the agent's coaching output or behavior.
Ingestion points: The skill pulls reply_text from email campaigns (Step 2) and full transcript strings from call recordings (Step 3) into the active context for reasoning.
Boundary markers: The instructions do not define clear delimiters (e.g., XML tags or unique markers) or specific system-level warnings to distinguish between instructional logic and the raw communication data being analyzed.
Capability inventory: The skill utilizes data-analysis, call-analysis, email-analysis, and reporting capabilities. While these are analytical, an injection could attempt to influence the 'Personalized Playbook' or 'Coaching Recommendations' to provide biased or harmful advice.
Sanitization: No sanitization, filtering, or instruction-extraction prevention measures are specified for the incoming email or transcript text.

sales-coaching