Asbury AI Solutions
Proof of Work · Model Context Protocol

Real Gateway.
Real Tools.
Real Configs.

MCP connectors for the Asbury Agentic OS. A live gateway serving twelve real org tools, with configs you can copy. Pick your AI tool, paste the config, connect in under a minute.

How It Works

What Is MCP?

The Model Context Protocol is an open standard — think of it as USB-C for AI tools. It replaces dozens of custom SDKs and one-off integrations with a single, universal protocol that any MCP-compatible AI can use.

One Gateway

The Asbury MCP Gateway at localhost:8811 — the local default — routes all tool requests through a single endpoint. One URL to remember, one config to share.

Tool Discovery

AI tools call tools/list to discover what's available. The gateway advertises its twelve live org tools — inbox, memory, health, kernel, cron, listeners — nothing invented.

Tool Execution

Each tool is a named function with typed parameters. The AI calls tools/call with arguments, the gateway dispatches to the right agent, and returns structured results.

Resource Access

Tools can read and write resources — files, database records, configuration, live data streams. Everything is a resource, and every resource has a URI.

Available Connectors

Tool Bridges

Six client configs, one gateway. The gateway serves the same twelve live org tools to every MCP client — inbox, memory, health, kernel, cron, listeners. No per-tool magic, no custom plugins. Click a card for the config.

v0

v0.dev

Coming +

Vercel's AI web app generator. v0 doesn't act as an MCP client today — when it does, the Asbury gateway will be a first-class target.

  • React
  • UI Gen
  • shadcn/ui
  • Roadmap

Status: v0 has no MCP client integration yet. Tracked on the roadmap — no config to ship until it lands.

CC

Claude Code

Live +

Anthropic's terminal-native coding agent. Add the gateway to your Claude Code config and every session can call live org tools — inbox, memory, health, kernel status — alongside its normal workflow.

  • Anthropic
  • Terminal
  • CLAUDE.md
  • MCP GA

Config: Add the MCP server entry to ~/.claude/settings.json (see the generator below). Every Claude Code session gets the gateway's twelve live org tools.

Cu

Cursor

Live +

Anysphere's AI-first code editor. Wire the gateway into .cursor/mcp.json and Cursor's agents get the same live org toolset, no plugin required.

  • Editor
  • Agent Mode
  • Multi-file
  • MCP GA

Config: Create .cursor/mcp.json in your project root. Cursor loads it on startup — the gateway's twelve live org tools appear in Composer and agent mode.

GC

GitHub Copilot

Beta +

GitHub's AI pair programmer. Drop the gateway into your repo's copilot-mcp config and Copilot chat reaches the same live org tools as everything else you run.

  • GitHub
  • VS Code
  • Chat
  • Public Preview

Config: Add to .github/copilot-mcp.json (top-level servers key — Copilot differs from the rest). Copilot loads MCP servers from this config.

Ws

Windsurf

Live +

Codeium's AI-native IDE. Point it at the gateway and its agents call the same twelve live org tools as your other clients.

  • Codeium
  • Flow
  • Cascade
  • MCP

Config: Configure in .windsurf/mcp_config.json (project) or ~/.codeium/windsurf/mcp_config.json (global) — not config.json.

Cx

Codex CLI

Live +

OpenAI's terminal coding agent. Connect via codex mcp add and it can reach the gateway's org tools from the shell.

  • OpenAI
  • Terminal
  • Sandbox
  • MCP

Config: codex mcp add asbury --transport http --url http://127.0.0.1:8811/mcp — or add [mcp_servers.asbury] to ~/.codex/config.toml (see the generator below).

DM

Dealership MCP

Live +

Dealership inventory API for commercial-truck sites — get_inventory, get_listing_details, search_inventory, health. Runs at :8702. This one is our own server, not a client config — the pattern in production.

  • Dealership
  • Inventory
  • Commercial Trucks
  • vAuto

Available tools: get_inventory, get_listing_details, search_inventory, health

Setup: Server runs on port 8702. Configure your MCP client to connect directly to http://127.0.0.1:8702/mcp. No auth required on loopback.

Copy-Paste Setup

Config Generator

Pick your AI tool and get the exact JSON config to plug into the Asbury MCP Gateway. Copy. Paste. Done.

Select a tool to generate config…

Auth: The gateway requires a bearer token. The config shows you exactly where it goes — the token itself never ships in this page's code. Export ASBURY_MCP_TOKEN in your shell and reference it via ${ASBURY_MCP_TOKEN}. Never paste your real token into this page. Configs point at the local gateway defaults — see the panel above.

Recorded, Not Live

Walk Through the Gateway

This is a real MCP session — initialize, tools/list, tools/call — captured from the live gateway and played back here, statically. No live calls from this page, no setup, no credentials. What you're reading is exactly what the gateway answered, payload for payload.

initialize Captured from the live gateway — 2026-08-05
Loading captured session…

A public page can't safely reach our loopback gateway — and we're not about to expose it without a fight. So you get the real thing, recorded.

Status

Roadmap

What's next for the Asbury MCP ecosystem.

  • Plugin SDK — Build your own MCP tools as pluggable Python packages Planned
  • Remote Gateway — Expose the gateway over Tailscale for remote tool access Beta
  • Auth & Access Control — Bearer-token auth on every request, loopback-only binding. Per-tool permissions and audit logging are future work. Shipped
  • Webhook Tools — Trigger n8n workflows and external webhooks from any AI tool Planned
  • Multi-Gateway Federation — Chain gateways across projects and machines Planned

Your Tool Isn't on This Page?

This page documents the gateway we actually run — every connector on it points at live infrastructure. If your stack isn't covered, we'll build the connector. You're not buying a pitch deck; you're buying the thing this page shows working.

Get in touch →