MCP Setup on Windows
Connect Agent-CoreX to your IDE or AI client on Windows using the Model Context Protocol (MCP). Once connected, your AI agent can call retrieve_tools to fetch relevant tools without passing hundreds of schemas in every request.
Last updated: April 2026
Prerequisites
- •An Agent-CoreX account and an API key — see Create an API Key
- •At least one of: VS Code, Cursor, Claude Code CLI, Windsurf, Cline, or Roo Code installed
- •For STDIO-based clients (Windsurf, Cline, Roo Code): Python + the
agent-corexCLI installed (pip install agent-corex)
Your MCP URL
Agent-CoreX exposes an HTTP SSE endpoint. Your personal MCP URL includes your API key as a query parameter:
https://api.agent-corex.com/mcp/sse?api_key=acx_your_api_key_hereThe easiest way to copy this URL is via Dashboard → Connect. The modal pre-fills your API key into the URL and provides 1-click install buttons for VS Code and Cursor.
VS Code
VS Code supports 1-click MCP install via a deep link or manual JSON config.
Option A — 1-click install
In the Agent-CoreX dashboard, go to Connect → Humans tab → VS Code. Click the "1-click install" button. VS Code opens and automatically adds the MCP server to your workspace.
Option B — Manual config (.vscode/mcp.json)
Create or edit .vscode/mcp.json in your workspace root:
{
"mcpServers": {
"agent-corex": {
"url": "https://api.agent-corex.com/mcp/sse?api_key=acx_your_api_key_here"
}
}
}Replace acx_your_api_key_here with your actual API key. Reload VS Code to apply.
Cursor
Cursor supports 1-click install or manual config in %USERPROFILE%\.cursor\mcp.json.
Option A — 1-click install
In the Agent-CoreX dashboard, go to Connect → Humans tab → Cursor. Click the "1-click install" button.
Option B — Manual config
Edit %USERPROFILE%\.cursor\mcp.json (create it if it doesn't exist):
{
"mcpServers": {
"agent-corex": {
"url": "https://api.agent-corex.com/mcp/sse?api_key=acx_your_api_key_here"
}
}
}Claude Code
Add Agent-CoreX to Claude Code via the terminal using the claude mcp add command:
claude mcp add --transport sse agent-corex "https://api.agent-corex.com/mcp/sse?api_key=acx_your_api_key_here"Claude Code will register the server under the name agent-corex. Verify with:
claude mcp listWindsurf
Windsurf uses a STDIO-based config. Edit %USERPROFILE%\.codeium\windsurf\mcp_config.json:
{
"mcpServers": {
"agent-corex": {
"command": "agent-corex",
"args": [
"serve"
],
"env": {
"AGENT_COREX_API_KEY": "acx_your_api_key_here"
}
}
}
}Requires the agent-corex CLI on your PATH (pip install agent-corex). Restart Windsurf after saving.
Cline
In the Cline VS Code extension, open MCP Servers in the Cline sidebar and paste the following config:
{
"mcpServers": {
"agent-corex": {
"command": "agent-corex",
"args": [
"serve"
],
"env": {
"AGENT_COREX_API_KEY": "acx_your_api_key_here"
}
}
}
}Roo Code
In Roo Code settings, navigate to MCP → Add server and paste:
{
"mcpServers": {
"agent-corex": {
"command": "agent-corex",
"args": [
"serve"
],
"env": {
"AGENT_COREX_API_KEY": "acx_your_api_key_here"
}
}
}
}Other clients
Continue
Add to Continue config.json under "mcpServers":
{
"name": "agent-corex",
"command": "agent-corex",
"args": [
"serve"
],
"env": {
"AGENT_COREX_API_KEY": "acx_your_api_key_here"
}
}Gemini CLI
Add to Gemini CLI MCP settings:
{
"mcpServers": {
"agent-corex": {
"url": "https://api.agent-corex.com/mcp/sse?api_key=acx_your_api_key_here"
}
}
}Amazon Q
Add to Amazon Q MCP configuration:
{
"mcpServers": {
"agent-corex": {
"url": "https://api.agent-corex.com/mcp/sse?api_key=acx_your_api_key_here"
}
}
}API / Custom SSE
Connect any SSE-capable client directly:
https://api.agent-corex.com/mcp/sse?api_key=acx_your_api_key_hereVerify your setup
Once your IDE is connected, make a query and confirm it appears in your usage dashboard. Check Queries Used →