AVA MCP Server

How to use

Connect your AVA client to the MCP server by pointing it to the stdio-based MCP entry. You can run the server locally in development to verify behavior, and then provide the client with the exact server configuration to establish ongoing communication.

In development you can start the server with the provided runtime command to simulate the MCP environment. This runs the server in dev mode and exposes the AVA toolset for testing.

To wire AVA to the MCP server, use the following example configuration. This demonstrates how AVA would load the stdio MCP server and execute the underlying Python script via a Windows Subsystem for Linux (WSL) command. Use your own paths and script names as needed.

{
  "mcpServers": {
    "AVA": {
      "command": "wsl.exe",
      "args": [
        "/home/ayoub/.local/bin/uv",
        "--directory",
        "/home/ayoub/Personal/MCP/1/src",
        "run",
        "mcp-server-example.py"
      ]
    }
  }
}

How to install

Follow these concrete steps to set up and run the MCP server locally. You will install the required runtime, start the server in dev mode, and connect an AVA client with the provided configuration.

Prerequisites you need before starting:

- Ensure you have Python installed if your server script depends on it.

- Install the UV tool to run MCP servers in development.

Proceed with these commands exactly as shown to set up and run in development mode.

# Install uv if you haven't already
curl -LsSf https://astral.sh/uv/install.sh | sh

# Test the server in dev mode
uv run mcp dev mcp-server-example.py

Next, add the MCP server configuration to your AVA client setup. Use the same path and script names, adapting them to your environment.

{
  "mcpServers": {
    "AVA": {
      "command": "wsl.exe",
      "args": [
        "/home/ayoub/.local/bin/uv",
        "--directory",
        "/home/ayoub/Personal/MCP/1/src",
        "run",
        "mcp-server-example.py"
      ]
    }
  }
}

Environment and security notes

Set up a local environment file to hold sensitive values securely and avoid committing them to version control.

.env
```
Add the following keys with your actual values.

The environment file should include credentials and tokens as shown in the example, and you should keep this file out of version control.

Also create a dedicated directory for agent credentials and tokens used by the Google OAuth flow, and keep them private.

Google OAuth setup and signing in

To allow the AVA to access Google services, follow the OAuth setup and sign-in flow. This creates a token that the server uses to access your account securely.

1) Create the required directory structure for storing credentials and tokens.

mkdir -p .config/ava-agent

2) Create a Google Cloud project, enable Gmail API, and configure OAuth consent screen with the necessary scopes.

3) Create desktop-style OAuth credentials and download the JSON file as credentials.json into the .config/ava-agent directory.

4) Initiate the OAuth flow to obtain a token if token.json is not present.

uv run oauth.py

Security notes

Protect sensitive files by excluding them from version control. Keep credentials and tokens secure and accessible only to the server runtime.

.gitignore
.env
.config/ava-agent/token.json
.config/ava-agent/credentials.json