outlook
Pass
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: SAFENO_CODEPROMPT_INJECTION
Full Analysis
- [NO_CODE]: The main functional script 'outlook' described in the README and SKILL.md is not included in the skill package, preventing a full security audit of the actual implementation logic.
- [PROMPT_INJECTION]: The skill's ability to retrieve and display external email content creates a surface for indirect prompt injection. Malicious emails containing hidden instructions could influence the AI agent when it processes retrieved messages. 1. Ingestion points: Email subject and body content retrieved via the 'read', 'list', and 'search' commands. 2. Boundary markers: None mentioned in the documentation or user instructions to delimit external content. 3. Capability inventory: Network operations (Graph API), local file system access for configuration and token storage, and potential subprocess execution of the CLI script. 4. Sanitization: No sanitization or filtering of external email content is described in the provided documentation.
Audit Metadata