diplomatic-negotiator

Pass

Audited by Gen Agent Trust Hub on Oct 5, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill instructions emphasize evidence-based analysis, professional boundaries, and safety protocols. It explicitly prohibits the fabrication of facts and the impersonation of individuals or institutions.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied information and external data sources for analysis, which represents a potential attack surface for indirect prompt injection.
  • Ingestion points: External data and user inputs described in the 'Required Input' section of SKILL.md.
  • Boundary markers: The 'Security and Ethics' section contains explicit directives to not treat third-party content as trusted instructions and to guard against prompt injection attacks.
  • Capability inventory: The skill is limited to read-only operations using tools like Read, Grep, Glob, and WebSearch, with no write or execution permissions.
  • Sanitization: Guidelines require the agent to cross-verify all information and clearly label the confidence level of conclusions (e.g., 'verified' vs 'pending confirmation').
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 5, 2026, 04:34 PM