Factifai

You run a Model Context Protocol (MCP) server to integrate Factifai with any MCP-capable tool. This server lets you create tests asynchronously and retrieve their results, enabling you to coordinate AI experiments and validation across your tooling stack with a consistent MCP interface.

How to use

To use the Factifai MCP Server, you run the local MCP server process and connect your MCP client to it. You will provide your chosen model provider (Bedrock or OpenAI) and the necessary credentials or access keys. Once started, you can initiate a test and later fetch the test results using the Tools exposed by this MCP server. The flow is designed to be tool-agnostic, so you can plug it into any MCP client that supports the MCP protocol.

How to install

Prerequisites you need before installing: Node.js version 16 or higher. You may also want an MCP client installed ( Hai Build, Cursor, Windsurf, Claude Desktop, or any compatible MCP client) to start using the server.

Install the latest version of the Factifai MCP Server using npx. Run these commands one by one in your terminal.

# Latest version
npx --yes @presidio-dev/factifai-mcp-server@latest

# Specific version
npx --yes @presidio-dev/factifai-mcp-server@1.2.3

Configuration

Configure the MCP server to run with your preferred model provider and credentials. The following environment variables are supported: MODEL_PROVIDER, OPENAI_API_KEY, AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, AWS_DEFAULT_REGION. You can choose either Bedrock or OpenAI as the model provider.

{
  "factifai": {
    "command": "npx",
    "args": ["--yes", "@presidio-dev/factifai-mcp-server@latest"],
    "env": {
      "MODEL_PROVIDER": "bedrock|openai",
      "OPENAI_API_KEY": "<your-openai-api-key>",
      "AWS_ACCESS_KEY_ID": "<your-aws-access-key-id>",
      "AWS_SECRET_ACCESS_KEY": "<your-aws-secret-access-key>",
      "AWS_DEFAULT_REGION": "<your-aws-region>"
    },
    "disabled": false,
    "autoApprove": []
  }
}

Model provider configuration examples

Choose your model provider and set the required credentials. Below are configuration examples for Bedrock and OpenAI.

// Bedrock Configuration Example
{
  "factifai": {
    "command": "npx",
    "args": ["--yes", "@presidio-dev/factifai-mcp-server@latest"],
    "env": {
      "MODEL_PROVIDER": "bedrock",
      "AWS_ACCESS_KEY_ID": "<your-aws-access-key-id>",
      "AWS_SECRET_ACCESS_KEY": "<your-aws-secret-access-key>",
      "AWS_DEFAULT_REGION": "<your-aws-region>"
    },
    "disabled": false,
    "autoApprove": []
  }
}

// OpenAI Configuration Example
{
  "factifai": {
    "command": "npx",
    "args": ["--yes", "@presidio-dev/factifai-mcp-server@latest"],
    "env": {
      "MODEL_PROVIDER": "openai",
      "OPENAI_API_KEY": "<your-openai-api-key>"
    },
    "disabled": false,
    "autoApprove": []
  }
}
```
## Factifai MCP integration with popular IDE and extension

You can integrate Factifai MCP Server with IDEs and extensions by configuring an MCP server entry in your IDE, using the standard MCP command and environment variables. The examples below show common IDE integrations and the required environment variables.

## Security

Protect your API keys and credentials. Use secure storage for OPENAI\_API\_KEY, AWS keys, and any other sensitive values. Rotate keys regularly and apply the principle of least privilege to your credentials.

## License

This project is licensed under the MIT License. You are free to use, modify, and distribute it in accordance with the license terms.

## Available tools

### testWithFactifai

Start a test with Factifai

### getFactifaiSessionResult

Get test result for a Factifai session

### listFactifaiSessions

List all Factifai test sessions