zoho_books-automation

Fail

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: HIGHPROMPT_INJECTIONREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION] (HIGH): The skill is highly susceptible to Indirect Prompt Injection (Category 8) because it ingests untrusted data from an external API and possesses significant side-effect capabilities.
  • Ingestion points: Data retrieved from Zoho Books via RUBE_MULTI_EXECUTE_TOOL (e.g., invoice descriptions, contact names, expense notes).
  • Boundary markers: Absent. The skill provides no instructions to the agent on how to delimit or ignore instructions embedded in the external data.
  • Capability inventory: The skill can execute write operations (create invoices, payments) and run code via RUBE_REMOTE_WORKBENCH.
  • Sanitization: None detected. Data from the API is used directly to drive subsequent agent decisions and tool calls.
  • [REMOTE_CODE_EXECUTION] (MEDIUM): The skill explicitly promotes the use of RUBE_REMOTE_WORKBENCH to execute Python code (e.g., using ThreadPoolExecutor) for batch processing. While a feature of the toolkit, this dynamic execution environment could be exploited if the agent interpolates untrusted data from Zoho Books into the scripts it generates for the workbench.
  • [EXTERNAL_DOWNLOADS] (LOW): The skill requires connecting to an external MCP server at https://rube.app/mcp. Per the security policy, this is flagged as an external dependency because the domain is not on the predefined list of trusted sources.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 17, 2026, 12:33 PM