I build the unglamorous layer underneath AI agents β
the tooling, gateways, and guardrails that keep them running long after the demo ends.
Currently focused on MCP developer tooling and agent-to-agent infrastructure.
A developer toolkit for MCP servers β inspect, test, and debug them before they break in production.
Ships as a CLI (inspector + test runner), a VS Code extension, and an agent-facing MCP server for automated validation.
A security gateway for agent-to-agent (A2A) communication β
because the moment agents start talking to other agents, you need a layer that doesn't trust either of them.
- LinkedIn: raeseoklee
- GitHub: raeseoklee





