Run multiple isolated MCP servers together in a single E2B sandbox, powered by Groq's ultra-fast inference.
E2B's MCP Gateway provides a unified interface to run multiple Model Context Protocol (MCP) servers inside isolated sandboxes. Instead of connecting to each MCP server separately, E2B creates a single gateway endpoint that routes requests to all configured MCPs—giving your AI access to 200+ tools through one connection.
- Unified Gateway - One endpoint for multiple MCP servers (Exa, Browserbase, Notion, etc.)
- Isolated Sandboxes - Each workflow runs in its own secure Linux environment
- Combined Capabilities - Mix search, browser automation, data storage, and more
- Type-Safe - Full TypeScript/Python SDK support
- Fast Setup - Configure multiple MCPs in seconds
- Sign up for services:
- E2B - Get API key for sandbox management
- Browserbase - Get API key + Project ID for browser automation
- Exa - Get API key for web search
- Groq - Get API key for LLM inference
All services have a free tier that is enough for this demo.
-
Install dependencies:
pip install e2b openai
-
Set environment variables:
export E2B_API_KEY='your_e2b_key' export BROWSERBASE_API_KEY='your_browserbase_key' export BROWSERBASE_PROJECT_ID='your_project_id' export EXA_API_KEY='your_exa_key' export GROQ_API_KEY='your_groq_key'
-
Run the notebook - Open
mcp-e2b.ipynbto see it in action
- Search for information with Exa
- Visit source websites with Browserbase to verify
- Extract and structure data for analysis
- Find companies with Exa
- Browse to their websites with Browserbase
- Extract contact info and pricing details
- Search for competitors with Exa
- Monitor their sites with Browserbase
- Track changes over time
- Research topics with Exa
- Screenshot examples with Browserbase
- Compile findings into reports
web_search_exa- AI-powered web search for current informationcompany_research- Research companies by crawling official websitescrawling- Extract content from specific URLslinkedin_search- Search LinkedIn for companies and peopledeep_researcher_start- Start comprehensive research reportdeep_researcher_check- Get completed research results
browserbase_stagehand_navigate- Navigate to URLsbrowserbase_stagehand_act- Perform actions with natural languagebrowserbase_stagehand_extract- Extract page contentbrowserbase_stagehand_observe- Find interactive elementsbrowserbase_screenshot- Capture page screenshotsbrowserbase_session_create- Create browser sessionsbrowserbase_session_close- Close browser sessions
E2B's Advantage: Instead of managing separate connections to Exa's MCP, Browserbase's MCP, and others, E2B creates one gateway that hosts them all. Your AI agent can seamlessly use web search, browser automation, and any other MCP tool through a single endpoint.
Groq's Speed: With up to 500+ tokens/second, Groq handles tool orchestration and response generation in milliseconds—making multi-step research workflows feel instant.
- E2B Homepage: e2b.dev
- E2B MCP Gateway Docs: e2b.dev/docs/mcp
- E2B MCP Quickstart: e2b.dev/docs/mcp/quickstart
- E2B Available MCP Servers: e2b.dev/docs/mcp/available-servers
- E2B Custom Servers: e2b.dev/docs/mcp/custom-servers
- E2B GitHub: github.com/e2b-dev/e2b
- Exa MCP Documentation: docs.exa.ai/reference/exa-mcp
- Exa MCP GitHub: github.com/exa-labs/exa-mcp-server
- Browserbase MCP Server: hub.docker.com/r/mcp/browserbase
- Stagehand Documentation: docs.stagehand.dev
- Groq API Documentation: console.groq.com/docs
- Groq Models: console.groq.com/docs/models
- Model Context Protocol: modelcontextprotocol.io
- MCP Specification: spec.modelcontextprotocol.io