centralstationcrm-automation

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [Indirect Prompt Injection] (HIGH): The skill is designed to ingest and process data from an external CRM (Centralstationcrm).
  • Ingestion points: Untrusted data enters the agent context via CRM record lookups and tool outputs.
  • Boundary markers: None are present; the instructions do not suggest using delimiters or 'ignore' instructions for external content.
  • Capability inventory: High-privilege actions are available, including RUBE_MULTI_EXECUTE_TOOL (for write operations) and RUBE_REMOTE_WORKBENCH (for remote code/tool execution).
  • Sanitization: No evidence of sanitization or schema validation for the data returned from the CRM.
  • [External Downloads / Dependencies] (MEDIUM): The skill requires the user to add an external MCP server endpoint (https://rube.app/mcp). This introduces a dependency on third-party infrastructure that manages the connection to the CRM toolkit.
  • [Command Execution] (MEDIUM): The use of RUBE_REMOTE_WORKBENCH with run_composio_tool() suggests the ability to execute complex, multi-step operations in a remote environment, which may include code execution or arbitrary tool sequences depending on the toolkit's underlying capabilities.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 07:47 AM