referral-dm
Pass
Audited by Gen Agent Trust Hub on Nov 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill operates as a specialized drafting assistant for professional communication. It includes explicit guardrails to prevent unauthorized automated actions, such as requiring user confirmation for any tool-based messaging and prohibiting mass automation or scraping in accordance with platform terms of service. No patterns of data exfiltration, credential harvesting, or malicious command execution were detected.
- [INDIRECT_PROMPT_INJECTION]: The skill provides an interface to process untrusted external data in the form of pasted hiring posts. While this represents an ingestion surface for indirect prompt injection (where a hiring post could contain instructions to influence the agent's output), the risk is mitigated by the skill's primary function as a drafting tool for human review and the absence of high-privilege automated output capabilities.
- Ingestion points: Section 0 (Inputs) instructs the agent to accept pasted hiring posts or job links provided by the user.
- Boundary markers: The instructions do not define specific delimiters for external content, relying on the agent's internal reasoning to differentiate between instructions and the pasted post content.
- Capability inventory: The skill reads local data from the
cracked-profileandcompany-inteltools and appends text to a local tracker file (outreach.md). It drafts text but does not autonomously execute network operations. - Sanitization: No specific sanitization or filtering of the pasted content is performed before processing.
Audit Metadata