Every Copilot user knows the tax: hand-wiring which MCP servers, skills, and agents each task needs, then watching them eat your context window whether you use them or not. GitHub Agent Finder, live since June 17 on all Copilot plans, hands that job back to Copilot itself.
## How it works
Describe a task in plain language. Agent Finder searches an index of available AI resources — MCP servers, skills, canvases, agents, tools — and returns ranked matches Copilot can pull in on demand, from whatever catalog you point it at. Agents load what the work calls for instead of carrying every tool just in case. Nothing gets silently connected; you approve what actually gets wired in.
## Why it matters
It runs on the open Agentic Resource Discovery spec, built with Google, Hugging Face, GoDaddy, and Microsoft — a rare cross-vendor agreement on how agents find tools at runtime. Enterprises define what’s discoverable in the same place they already govern Copilot, so Agent Finder only ever surfaces permitted resources. As MCP servers and skills multiply into the thousands, discovery, not raw capability, becomes the bottleneck — and that is exactly what this targets.

Leave a comment