Quick Start: Setting Up Your First MCP Server

​

Before You Start

Before you begin, make sure you have:

• ☑️ VS Code installed (version 1.99 or newer)
• ☑️ GitHub Copilot installed and configured in VS Code
• ☑️ Python installed on your computer (version 3.8 or newer) - Download Python
• ☑️ Your Elitea Project ID (find this in Elitea Settings → Configuration)
• ☑️ Your Elitea Authentication Token (create one in Elitea Settings → Personal Tokens)
• ☑️ Your Elitea deployment URL (example: https://next.elitea.ai)

---

​

What is MCP and Why Do I Need It?

MCP stands for Model Context Protocol. It’s a way for different AI tools to talk to each other. Why it matters to you: By setting up an MCP server, you can use your Elitea agents directly inside VS Code while you code. No need to switch windows or copy-paste between apps.

---

​

Step-by-Step Setup

​

Part 1: Install the Alita MCP Client

The “client” is a small program that handles communication between VS Code and Elitea.

1. Open PowerShell
   • Press Windows key, type “PowerShell”, and press Enter
   • This opens a window where you’ll type commands
2. Install pipx (a tool that helps install Python programs)
   • Copy this command and paste it into PowerShell, then press Enter:
     python -m pip install --user pipx
     
   • Wait for it to finish
   
3. Set up pipx’s path
   • This makes sure your computer can find pipx. Copy and run:
     python -m pipx ensurepath
     
   
4. Install alita-mcp
   • This is the actual client program. Copy and run:
     pipx install alita-mcp
     
   • Wait for installation to complete
   
5. Close and reopen PowerShell
   • This refreshes your environment so the new program is recognized

---

​

Part 2: Configure the Client

Now you’ll tell the client how to connect to your Elitea account.

1. Run the setup wizard
   • In PowerShell, type:
     alita-mcp bootstrap
     
   • Press Enter
2. Answer the prompts
   • Deployment URL: Type your Elitea URL (example: https://next.elitea.ai) and press Enter
   • Authentication token: Paste your token (the one you copied earlier) and press Enter
   • Host: Just press Enter to accept the default (0.0.0.0)
   • Port: Just press Enter to accept the default (8000)
   • Project ID: You can either:
     • Enter your Project ID number now (find it in Elitea Settings → Configuration), OR
     • Press Ctrl+C to skip this step and add the Project ID later in VS Code (see Part 4)
3. Verify success
   • You should see a message confirming the configuration was saved
   • If you see errors, double-check your URL and token

---

​

Part 3: Connect to VS Code

Now you’ll add the MCP server to VS Code so it can talk to Elitea.

1. Open VS Code
   • Launch the VS Code application
2. Open your project
   • Open the folder or workspace where you want to use Elitea agents
3. Open GitHub Copilot Chat
   • Click the chat icon in the left sidebar (or press Ctrl+Alt+I)
   
4. Switch to Agent mode
   • In the chat panel, look for “Agent mode” and click it
   • This enables the use of tools (like your Elitea agents)
5. Add the MCP server
   • Click the Select Tools icon (looks like a wrench or tools)
   • Click Add MCP Server icon
   • Click + Add MCP Server
   
   • Select Command stdio
   
   • For “Command”, type: alita-mcp and press Enter
   
   • For “Server ID”, use the suggested name or type something like elitea
   
6. Choose where to save the settings
   • You’ll see two options:
     • Workspace – Only works in this project folder
     • Global – Works in all your VS Code projects
   • Choose based on your needs (Workspace is good for team projects)
   

---

​

Part 4: Add Your Project ID

This tells the server which Elitea project to connect to.

​

Option A: If You Chose Workspace

1. The config file opens automatically
   • After selecting Workspace, VS Code will automatically open .vscode/mcp.json
   • If it doesn’t open automatically, look for the file at .vscode/mcp.json and click to open it
2. Add your project ID
   • Find the line that says "args": []
   • Replace it with:
     "args": ["run", "--project_id", "YOUR_PROJECT_ID"]
     
   • Replace YOUR_PROJECT_ID with your actual number (example: 25)
   
3. Save the file
   • Press Ctrl+S
4. Start the server
   • In Copilot Chat, click the Start button
   

---

​

Option B: If You Chose Global

1. The settings file opens automatically
   • After selecting Global, VS Code will automatically open your global mcp.json configuration file
   • If it doesn’t open automatically, look for a new tab in the VS Code editor with the file
2. Add your project ID
   • Find the MCP server section you just added
   • Find the line that says "args": []
   • Replace it with:
     "args": ["run", "--project_id", "YOUR_PROJECT_ID"]
     
   • Replace YOUR_PROJECT_ID with your actual number
   
3. Save the file
   • Press Ctrl+S
4. Start the server
   • In Copilot Chat, click the Start button
   

---

​

Part 5: Tag Your Agents

For an agent to be available as a tool, it must be tagged in Elitea.

​

Tagging Agents:

1. Log into Elitea
   • Open your Elitea deployment in a web browser
2. Find your agent
   • Navigate to the agent or pipeline you want to use
3. Add the “mcp” tag
   • Look for the tags section
   • Add a tag named mcp (all lowercase)
   • Save your changes
   

​

Complete Setup:

4. Restart the MCP server in VS Code
   • Go back to VS Code
   • In Copilot Chat, stop the server and start it again
   • This syncs the newly tagged agents

---

​

Part 6: Use Your Agent

Now you’re ready to use your Elitea agent from VS Code!

1. Open Copilot Chat
   • Make sure you’re in Agent mode
2. Click the Tools icon
   • You should see your tagged agents listed
   
3. Select the tools you want to use
   • All tools are selected by default
   • You can deselect any you don’t need right now
4. Type your request
   • In the chat box, type # followed by your agent’s name
   • Example: #BA Assistant Show me the scope of PLAN-3012
   
5. Run the tool
   • Click Run [tool name] when prompted
   
6. Confirm execution
   • The first time, you’ll be asked to confirm
   • Click Continue (you can choose to auto-confirm for future runs)
7. Review the results
   • Your agent will execute and return results directly in the chat
   

---

​

How to Know It’s Working

✔️ You’ve succeeded if:

• The MCP server shows as “Running” in VS Code
• Your tagged agents appear in the Tools list
• When you call an agent, you see output in the chat
• No error messages appear

✘ Something’s wrong if:

• You see “Command not found” errors → Reinstall the client
• No agents appear in the Tools list → Check that they’re tagged with mcp
• Server won’t start → Check your Project ID and authentication token

---

​

Common Problems & Fixes

​

Problem: “alita-mcp: command not found”

Fix: Make sure you completed all installation steps, including pipx ensurepath. Close and reopen PowerShell.

​

Problem: No agents showing up

Fix:

1. Make sure your agents are tagged with mcp in Elitea
2. Restart the MCP server in VS Code
3. Check you’re using the correct Project ID

​

Problem: “Error indicator” in VS Code

Fix:

1. Click the error in the Chat view
2. Select Show Output to see details
3. Look for specific error messages in the log

​

Problem: Server keeps disconnecting

Fix:

1. Verify your authentication token is still valid
2. Check your internet connection
3. Confirm your Elitea deployment URL is correct

​

Problem: Multiple alita-mcp versions installed (pip and pipx conflict)

Fix: If you see errors or unexpected behavior, you may have multiple versions of alita-mcp installed. This happens when you installed it with both pip and pipx.

1. Check for multiple installations:
   pip list | Select-String alita-mcp
   pipx list | Select-String alita-mcp
   
2. Remove the pip version (keep pipx):
   pip uninstall alita-mcp -y
   
3. Verify the correct version is active:
   where.exe alita-mcp
   
   You should see: C:\Users\<YourUsername>\.local\bin\alita-mcp.exe
4. Restart VS Code and the MCP server

---

​

What You Can Do Next

Now that your MCP server is running, you can:

• Call multiple agents in one conversation
• Use agents to analyze code, create documentation, or fetch data
• Automate repetitive tasks directly from VS Code
• Combine Elitea agents with other Copilot features

---

​

Advanced Options

---

---