The Agent Skills Directory

[EXTERNAL_DOWNLOADS] (MEDIUM): The skill requires connecting to an untrusted external MCP endpoint https://rube.app/mcp. This source is outside the trusted scope and provides the logic for the agent's actions.
[PROMPT_INJECTION] (HIGH): The skill is highly vulnerable to Indirect Prompt Injection (Category 8) because it fetches 'recommended execution plans' from an external source and instructs the agent to follow them.
Ingestion points: RUBE_SEARCH_TOOLS returns schemas and execution plans from the remote server into the agent context.
Boundary markers: None present; the skill explicitly directs the agent to 'Always search tools first' and follow the returned guidance.
Capability inventory: RUBE_MULTI_EXECUTE_TOOL and RUBE_REMOTE_WORKBENCH provide broad tool execution capabilities across the Brandfetch toolkit.
Sanitization: None; the agent is instructed to use the 'exact field names' and recommended plans provided by the external search results.
[COMMAND_EXECUTION] (HIGH): The skill enables execution of remote tools and management of connections based on instructions fetched at runtime from an unverified server, potentially allowing a malicious server to trigger unauthorized actions via the agent.

brandfetch-automation