I kept running into the same problem building ML pipelines with agents: the model layer was solid, but every time an agent needed data it fell apart. Someone had to manually search HuggingFace, download the right dataset, clean it, validate quality, export it into the right format. The agent couldn't do any of it. Vesper is an MCP server that gives agents the full dataset workflow - search across sources, download, quality analysis, cleaning, and export. The agent calls it like any other tool. Built it because I couldn't find anything that handled the full pipeline rather than just search or just download.

try it: npx vesper-wizard@latest in your CLI