manic light
Hi everyone,
I’m having an issue with the Hermes agent. I’ve set the model to Minimax 2.7 and I’m on the $10 Token plan. The model is supposed to support image recognition, but it doesn’t seem to work in Hermes. I get an error every time I try to use the vision feature.
Am I the only one experiencing this?
manic light
Minimax use mcp
Quick question: Is there a way to make native Hermes tools (like browser_vision or other built-in features) automatically leverage MCP functions like understand_image? Or do they always require manual triggering?
karmic dome
Not automatically, no.
Right now Hermes native tools like browser_vision and vision_analyze are their own built-in paths. MCP tools like MiniMax’s understand_image are a separate path.
So native Hermes tools do not automatically delegate to MCP tools behind the scenes. If MiniMax’s understand_image is available, Hermes would need to use it as an MCP tool explicitly, not as a transparent backend for the native vision tools.
So the practical answer is: they are separate unless someone adds explicit integration or router logic to connect them. For now, treat it as manual or explicit MCP-tool usage, not automatic leverage.