Local MCP Gateway
See, allow, and log every MCP tool call your agents make — one local policy across every editor you use.
This capability is on our roadmap — not shipped today. Early access opens to a small group at a time; requesting a spot is free and carries no commitment.
Request early accessThe problem
Every agent client configures MCP tools its own way — one file for Cursor, another for VS Code, another for your CLI agents — and none of them show you what those tools actually do once connected. The result is standing, all-access credentials scattered across editors, no single view of what your agents are calling, and no one place to say no. Governing agent tools should not require configuring every editor separately.
What you'd get
- One local gateway between your agents and their MCP tools — every call visible, allowed, or logged in one place.
- Least-privilege, revocable grants per tool and per agent — instead of standing, all-access tokens.
- One policy that follows you across Cursor, Claude Code, VS Code, and CLI agents — no per-editor configuration.
- Local-first: decisions happen on your machine, and your traffic never has to route through someone else's cloud.
Who it's for
- Developers running AI agents in more than one editor who want one place to govern them all.
- Platform engineers accountable for which MCP servers and tools agents may reach.
- Security teams that need an audit-ready record of every agent tool call.
Get early access to Local MCP Gateway
See, allow, and log every MCP tool call your agents make — one local policy with least-privilege grants, across Cursor, Claude Code, and VS Code.