draft-outreach
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- Indirect Prompt Injection Surface: The skill processes external, potentially untrusted content from emails, chat transcripts, and enrichment data, which creates a surface for indirect prompt injection attacks.
- Ingestion points: External data enters the agent's context through email thread history, CRM activity records, and third-party enrichment results (SKILL.md, Step 2).
- Boundary markers: The instructions explicitly define these sources as "untrusted content" and mandate they be treated as data rather than instructions, which serves as a defensive boundary.
- Capability inventory: The skill possesses the ability to draft and send emails, add contacts to sales sequences (Apollo, Outreach, Salesloft), and log CRM activity (SKILL.md, Step 4, 4b, 5).
- Sanitization: The skill implements sanitization measures by requiring the agent to show "content-originated" actions to the user before execution and by forbidding the rendering of links found within untrusted content.
Audit Metadata