Client integrations

EdgeRouteAI is OpenAI-compatible at api.edgerouteai.com/v1, and exposes MCP at mcp.edgerouteai.com. Most clients work with one config tweak. The Setup page generates these snippets pre-filled with your API key.

Claude Desktop

claude_desktop_config.json

{
  "mcpServers": {
    "edgerouteai": {
      "command": "npx",
      "args": ["-y", "@edgerouteai/mcp-stdio"],
      "env": { "EDGEROUTE_KEY": "edgrt_…" }
    }
  }
}

Cursor

Same MCP config in Cursor's MCP settings. EdgeRoute appears as a callable tool.

~/.cursor/mcp.json

{
  "mcpServers": {
    "edgerouteai": {
      "command": "npx",
      "args": ["-y", "@edgerouteai/mcp-stdio"],
      "env": { "EDGEROUTE_KEY": "edgrt_…" }
    }
  }
}

Cline (VS Code)

Cline reads cline_mcp_settings.json in the VS Code globalStorage. Same MCP shape.

Continue (VS Code)

Continue treats EdgeRoute as a custom OpenAI-compatible provider. Add to ~/.continue/config.json:

config.json

{
  "models": [
    {
      "title": "EdgeRouteAI auto",
      "provider": "openai",
      "model": "auto",
      "apiBase": "https://api.edgerouteai.com/v1",
      "apiKey": "edgrt_…"
    }
  ]
}

Aider

shell

export OPENAI_API_BASE=https://api.edgerouteai.com/v1
export OPENAI_API_KEY=edgrt_…
aider --model auto

Open WebUI / LibreChat / TypingMind / Chatbox / LobeChat

All five expose a "Custom OpenAI-compatible provider" or "Add base URL" setting. Plug in:

Base URL: https://api.edgerouteai.com/v1
API Key:  edgrt_…
Model:    auto    (or any of the 11 explicit models)

LangChain

Python

from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    model="auto",
    api_key="edgrt_…",
    base_url="https://api.edgerouteai.com/v1",
)

Vercel AI SDK

TypeScript

import { createOpenAI } from '@ai-sdk/openai'

const edgeroute = createOpenAI({
  apiKey: process.env.EDGEROUTE_KEY,
  baseURL: 'https://api.edgerouteai.com/v1',
})

const result = await streamText({
  model: edgeroute('auto'),
  prompt: 'Hello',
})