gog
Pass
Audited by Gen Agent Trust Hub on Jul 31, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions in
SKILL.mdandprompt.mdemphasize safety best practices, such as verifying object IDs before performing writes and using the narrowest possible scope for queries. - [INDIRECT_PROMPT_INJECTION]: The skill inherently processes untrusted data from external sources (e.g., Gmail threads and Google Docs) and possesses write capabilities (e.g., drafting emails and updating spreadsheets).
- Ingestion points: The agent reads data from Gmail, Drive, Docs, and Sheets as specified in
SKILL.md. - Boundary markers: The instructions do not explicitly implement delimiters or warnings to ignore instructions embedded within retrieved workspace content.
- Capability inventory: The agent can perform write operations including creating calendar events, drafting emails, and updating spreadsheet ranges (
prompt.md). - Sanitization: No explicit sanitization or filtering of external content is defined in the instructions.
- Note: This is a standard characteristic of tools designed to interact with communication and document platforms.
Audit Metadata