VegaMCP is a production-grade MCP server that runs an autonomous AI agent swarm with persistent memory, browser automation, multi-model reasoning, and a security gateway. It exposes a wide range of capabilities as tools within the MCP protocol, enabling clients to orchestrate agents, query tools, manage memory, and perform automated tasks across web, mobile, and API testing domains. The server is implemented in TypeScript/Node.js and is intended to be run via Node with the built build/index.js entry point. You can connect to VegaMCP from an MCP client by pointing to the server process and using the provided tool calls to execute the 65+ integrated capabilities (reduced into 15 high-level clusters in v7.0 for efficiency).

To use VegaMCP, start the server process with the appropriate environment variables (API keys for models and integrations). Typical clients will interact with the server through the MCP protocol, sending requests to initialize agent swarms, perform reasoning over contexts, run automated browser actions using Playwright-based tooling, and access security features like zero-trust identity and A2A protocol communications. The tooling set includes capabilities for web/app testing, code generation and evaluation, data retrieval from external sources, and orchestrating coordinated agent actions across multiple models and tools.

Prerequisites:

• Node.js 18+ (LTS)
• npm 9+ (or yarn)

Installation steps:

1. Clone the repository git clone https://github.com/Pastarafian/VegaMCP.git cd VegaMCP

2. Install dependencies npm install

3. Copy environment template and configure keys cp .env.example .env

   Edit .env and populate required API keys and settings

4. Build the project npm run build

5. Run the server (example)

   Ensure your environment is configured and paths are correct

   node build/index.js

Optional for VS Code integration:

• Create a mcp.json in your workspace as shown in the Quick Start example to connect the MCP client to VegaMCP.

Notes and tips:

• At least one reasoning model key is required (OPENROUTER_API_KEY, DEEPSEEK_API_KEY, KIMI_API_KEY). Fill as many as you need for richer capabilities.
• Configure budget controls via TOKEN_DAILY_BUDGET_USD and TOKEN_HOURLY_BUDGET_USD to manage token usage.
• VEGAMCP_TOOL_PROFILE can be set to full, minimal, research, coding, or ops to tailor tool usage and context consumption.
• Environment variables can be placed in .env or in the mcp.json env block when connecting via IDE integrations.
• If you encounter connectivity issues, verify that the built entry point path (build/index.js) matches your local build output and that Node.js 18+ is active in your shell.
• For VS Code integration, ensure the path in .vscode/mcp.json points to the built index and that the working directory is VegaMCP root.
• Regularly update dependencies and rebuild to pick up improvements in tool clusters and protocol handling.