managing-google-sheets
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- PROMPT_INJECTION (HIGH): The skill is vulnerable to Indirect Prompt Injection. It reads data from external Google Sheets using
read tableandread rangebut lacks boundary markers or instructions to treat cell content as data only. - Ingestion points:
sheets-cli read table,sheets-cli read range, andsheets-cli headerinSKILL.md. - Boundary markers: Absent. There are no delimiters or 'ignore instructions' warnings.
- Capability inventory: The skill has significant write capabilities, including
update key,update row,append, andbatchoperations. - Sanitization: Absent. No evidence of input validation or escaping for spreadsheet content.
- DATA_EXFILTRATION (MEDIUM): The skill facilitates the handling of sensitive authentication secrets. The
auth login --credentials <oauth-client.json>command explicitly directs the agent to interact with OAuth client secret files, creating a risk of credential exposure if the agent is compromised or tricked into leaking the file contents.
Recommendations
- AI detected serious security threats
Audit Metadata