wolfram-alpha-api-automation
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS] (MEDIUM): The skill instructs users to add an external MCP server from
https://rube.app/mcp. This source and the associated service provider (Composio/Rube) are not within the trusted entity list, making the dependency unverifiable. - [PROMPT_INJECTION] (HIGH): The skill has a significant attack surface for Indirect Prompt Injection. * Ingestion points: API responses from Wolfram Alpha processed through
RUBE_MULTI_EXECUTE_TOOL. * Boundary markers: Absent. There are no instructions to the agent to treat incoming data as untrusted or to use delimiters to prevent instruction hijacking. * Capability inventory:RUBE_MULTI_EXECUTE_TOOLfor multi-tool execution andRUBE_REMOTE_WORKBENCHfor remote operations. * Sanitization: Absent. The skill lacks validation or filtering of external API content before it is used in subsequent agent reasoning or tool calls. - [REMOTE_CODE_EXECUTION] (MEDIUM): The
RUBE_REMOTE_WORKBENCHtool allows the agent to orchestrate or execute remote tasks via an external provider's infrastructure, which introduces risks of arbitrary execution or data exposure if the middleman is compromised.
Recommendations
- AI detected serious security threats
Audit Metadata