kraken-io-automation

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS] (MEDIUM): The skill instructs users to add an external MCP server endpoint (https://rube.app/mcp). This source is not within the trusted domain list, and its behavior cannot be verified, posing a risk of remote tool manipulation or data redirection.- [COMMAND_EXECUTION] (HIGH): Through the use of RUBE_MULTI_EXECUTE_TOOL and RUBE_REMOTE_WORKBENCH, the skill provides capabilities to execute arbitrary tools and potentially remote code (via run_composio_tool()). This level of privilege is dangerous when combined with external data ingestion.- [PROMPT_INJECTION] (HIGH): Category 8: Indirect Prompt Injection surface identified.\n
  • Ingestion points: Untrusted data enters the agent context via responses from the Kraken IO toolkit (e.g., file metadata, API status messages, or optimized content data).\n
  • Boundary markers: None. The instructions do not provide delimiters or warnings for the agent to ignore instructions embedded in the tool outputs.\n
  • Capability inventory: RUBE_MULTI_EXECUTE_TOOL, RUBE_REMOTE_WORKBENCH (provides tool execution and workbench capabilities with side effects).\n
  • Sanitization: Absent. There is no logic for validating or sanitizing the content received from the Kraken IO service before it influences subsequent agent decisions or tool calls.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 12:56 PM