The Agent Skills Directory

Indirect Prompt Injection Surface: The skill is designed to ingest and analyze data from external sources, including emails, chat transcripts, and documents, which could contain unintended instructions.
Ingestion Points: Data is pulled from email bodies, CRM contact records, and uploaded book spreadsheets as detailed in SKILL.md.
Boundary Markers: The skill explicitly defines external content as untrusted and instructs the agent to treat it as data only, ignoring any embedded instructions.
Capability Inventory: The agent accesses CRM records and email threads to generate scoring and recommendations, with suggested actions requiring user confirmation.
Sanitization Measures: Robust rules are in place to prevent the rendering of links found within untrusted content and to ensure content-originated actions are shown to the user before execution.

account-tiering