connect-ai-googlesheets

Pass

Audited by Gen Agent Trust Hub on Aug 28, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of data from external Google Sheets, which creates a surface for indirect prompt injection if the sheet content contains malicious instructions that the agent might inadvertently execute.
  • Ingestion points: Data is read from dynamic tables named in the format [SpreadsheetName_SheetName] as described in the 'Query Process' section of SKILL.md.
  • Boundary markers: The instructions do not specify the use of delimiters or 'ignore embedded instructions' prompts when processing sheet cell data.
  • Capability inventory: The agent has the capability to perform SQL operations (SELECT, INSERT, UPDATE) and execute stored procedures such as CreateSpreadsheet, AddSheet, DeleteSheet, and DeleteSpreadsheet (SKILL.md).
  • Sanitization: The skill recommends using parameterization (@param or ?) for values in SQL statements, which helps mitigate SQL injection but does not prevent the LLM from obeying instructions found within the data itself.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 28, 2026, 02:09 PM
Security Audit — agent-trust-hub — connect-ai-googlesheets