The Agent Skills Directory

Indirect Prompt Injection Surface: The skill processes external, untrusted communication channels such as emails and meeting transcripts while maintaining capabilities to suggest CRM updates, outreach drafts, and meeting schedules.
Ingestion Points: External communication data, including emails, chat logs, and transcripts, are scanned during the deal analysis phase in SKILL.md.
Boundary Markers: The instructions explicitly mandate that external content must be treated strictly as data rather than instructions, requiring the agent to report any instruction-like text rather than executing it.
Capability Inventory: The skill proposes actions via update-opportunity, draft-outreach, and schedule-meeting tools based on the analyzed data.
Sanitization & Verification: Content-originated actions require explicit user confirmation showing recipients and content before execution, and the rendering of links found within untrusted text is prohibited.

deal-signals