Originally published at agentlair.dev . The agent-trust pattern that's been live at agentlair.dev had one rough edge. To verify another agent, you had to write an HTTP client. The verification logic ran in a worker. The consumer had to glue it in. That's now an npx command. @agentlair/mcp-server shipped to npm today. It's a Model Context Protocol server. Any MCP client (Claude Desktop, Cursor, Cline, Smithery) pulls it in and gains five tools that talk to AgentLair's verification endpoints directly. The agent verifies the agent. Three lines of config Open claude_desktop_config.json . On macOS that's ~/Library/Application Support/Claude/claude_desktop_config.json . Add: { "mcpServers" : { "agentlair" : { "command" : "npx" , "args" : [ "-y" , "@agentlair/mcp-server" ] } } } Enter fullscreen mode Exit fullscreen mode Restart Claude Desktop. Five new tools show up in the MCP panel. Type into chat: "verify the agent acc_qgdxSULsXsmtHklZ for me." Claude calls verify_agent , gets a score, and renders it.…