style-analyzer

Pass

Audited by Gen Agent Trust Hub on Aug 8, 2026

Risk Level: SAFEDATA_EXFILTRATION
Full Analysis
  • Broad Communication Access: The skill is designed to read up to 30 sent emails and hundreds of Teams messages across 25 different chats. While this is necessary for the stated purpose of voice mimicry, it involves significant access to the user's private communication history.
  • Storage of Communication Patterns: The skill distills personal writing habits (greetings, common phrases, quirks) and saves them to the agent's persistent memory. This creates a derived record of the user's communication style that persists across sessions and may be accessible by other skills.
  • Content Extraction Surface: Although the skill instructions explicitly state not to store verbatim message content, the analysis process requires the agent to process sensitive message bodies, which is a potential surface for indirect prompt injection if those messages contain malicious instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 8, 2026, 12:34 AM
Security Audit — agent-trust-hub — style-analyzer