assist-gmail

Pass

Audited by Gen Agent Trust Hub on Jun 29, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill processes external email content which may contain malicious instructions meant to manipulate the agent's behavior.
  • Ingestion points: Email data, including subjects, snippets, and AI summaries, are retrieved from the backend server via /api/google/emails and related endpoints in scripts/gmail-axi.mjs.
  • Boundary markers: The skill uses a custom structured format (TOON) to separate data fields, which provides some structural delimitation but does not include explicit warnings to the agent to disregard instructions found within the email content.
  • Capability inventory: The skill is capable of network communication (GET/POST) to a server and presenting data to the user. It does not have permissions to execute arbitrary system commands or access sensitive local files directly.
  • Sanitization: While the output is formatted as structured data, the raw content of the emails is passed through without specific sanitization for prompt injection patterns.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 29, 2026, 07:22 PM
Security Audit — agent-trust-hub — assist-gmail