google-sheets
Pass
Audited by Gen Agent Trust Hub on Apr 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it reads and processes data from external Google Sheets that could contain malicious instructions.\n
- Ingestion points: The agent reads the output of
scripts/read.rbduring the 'Analyze' step in the flow.\n - Boundary markers: The instructions do not define any delimiters or provide guidance to the agent to treat sheet content as untrusted data.\n
- Capability inventory: The skill provides tools for reading data (
read.rb), writing data (write.rb), appending rows (append.rb), and listing sheet metadata (list_sheets.rb).\n - Sanitization: No validation, escaping, or filtering is performed on the data retrieved from the API before it is presented to the agent for analysis.\n- [COMMAND_EXECUTION]: The
SKILL.mdinstructions guide the agent to execute shell commands with arguments derived from user input, such as the spreadsheet URL and the service account JSON key. Specifically, the commandruby save_key.rb 'PASTED_JSON_CONTENT'uses single quotes to wrap a potentially large and untrusted blob of data, which could lead to command injection if the content contains shell metacharacters and the agent platform does not sanitize the invocation.
Audit Metadata