kai-sales-meeting-prep

Pass

Audited by Gen Agent Trust Hub on Sep 28, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data such as sales development representative (SDR) replies and meeting transcripts.
  • Ingestion points: The skill ingests transcripts, call notes, and SDR replies as primary source material for analysis (SKILL.md).
  • Boundary markers: The instructions require the agent to explicitly label extracted data as prospect_said, seller_said, inferred, or missing to maintain source integrity.
  • Capability inventory: The skill has the capability to write Markdown files to the workspace/ directory and draft follow-up emails based on the processed content.
  • Sanitization: The skill includes defensive instructions to not treat transcript statements as verified facts unless the context is clear and to avoid mutating CRM or calendars without explicit approval.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 28, 2026, 04:41 PM