The error is on your screen in a terminal, a log viewer, or a canvas-rendered panel, and copying it out loses the context around it.
A capture is a first-class prompt in Omni. Captures go to vision-capable models as images, so a stack trace is readable whether it lives in the DOM, a canvas, an image, or an in-page PDF. You can ask what it means in the editor, or inside a side-panel conversation when you have several follow-up questions.
Take a visible-area or region capture of the trace, send it to a model, and ask. Use region capture for long traces so the text stays legible after the 1,568 pixel downscale, and make sure the image tier is pointed at a vision-capable model, since DeepSeek models are text only.