invoice-exception-manager-briefing

Pass

Audited by Gen Agent Trust Hub on Aug 4, 2026

Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
  • [PROMPT_INJECTION]: The skill incorporates external invoice identifiers and summaries into its call briefing, creating a potential surface for indirect prompt injection. Malicious instructions embedded in an invoice record could attempt to influence the agent's verbal output during the call. • Ingestion points: invoice-exception identifier and briefing summary (SKILL.md). • Boundary markers: The skill does not define specific technical delimiters or escaping for the injected summary text (SKILL.md). • Capability inventory: Outbound voice calls via the CALL-E service (SKILL.md). • Sanitization: Strong procedural controls including manager identity verification and redaction of payment/PII data (SKILL.md, safety.md).
  • [NO_CODE]: The skill consists entirely of instructional markdown and reference files, with no executable scripts or binaries included for analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 4, 2026, 10:05 AM
Security Audit — agent-trust-hub — invoice-exception-manager-briefing