The Agent Skills Directory

[EXTERNAL_DOWNLOADS] (HIGH): The skill requires the addition of an untrusted MCP server endpoint (https://rube.app/mcp). This source is not on the trusted external sources list, making the dependency unverifiable and potentially malicious.
[REMOTE_CODE_EXECUTION] (HIGH): The skill is designed to execute tools provided by the remote MCP server via RUBE_MULTI_EXECUTE_TOOL and RUBE_REMOTE_WORKBENCH. This creates a pathway for remote code execution if the external server provides tools with malicious side effects.
[PROMPT_INJECTION] (HIGH): Significant Indirect Prompt Injection surface (Category 8).
Ingestion points: Data enters the agent's context through RUBE_SEARCH_TOOLS, which returns external tool slugs, schemas, and 'recommended execution plans' from an untrusted remote server.
Boundary markers: Absent. There are no instructions to validate or sanitize the execution plans or tool definitions returned by the remote server.
Capability inventory: The skill has extensive execution capabilities, including RUBE_MULTI_EXECUTE_TOOL for running discovered tools and RUBE_REMOTE_WORKBENCH for remote operations.
Sanitization: Absent. The agent is explicitly told to use the returned schemas and plans directly to execute workflows, making it vulnerable to malicious instructions embedded in the search results.

flexisign-automation