The MCP Discoverability Crisis: Why 73% of Tools Remain Invisible to AI Agents

This research reveals a critical infrastructure problem in the AI agent ecosystem: most tools lack the metadata agents need to discover and use them autonomously. However, the fixes are surprisingly simple—adding action verbs, defining required fields, and including usage context can improve tool selection rates by 3.6x. For developers building agent-powered applications, this highlights both an urgent market gap and a competitive opportunity.