MCP Servers Mono-Repo (Emergence AI)

A mono-repository for hosting multiple Model Context Protocol (MCP) servers, each deployed as a separate Cloud Run service.

Available Servers

ChEMBL MCP Server

An MCP server that provides tool-based access to the ChEMBL chemical database via the public ChEMBL REST API.

Features:

• 27 MCP tools covering:
  • Compound search & retrieval
  • Target search & retrieval
  • Bioactivity and assay lookup
  • Similarity / substructure search
  • Basic property and "drug-likeness" summaries derived from ChEMBL molecule properties
• Resource templates (URI reads), including:
  • chembl://compound/{chembl_id}
  • chembl://target/{chembl_id}
  • chembl://assay/{chembl_id}
  • chembl://activity/{activity_id}
  • chembl://search/{query}

Location: src/chembl/

Prerequisites

• Node.js 20+
• pnpm 8+ (for workspace management)

Repository Structure

/
├── src/
│   ├── chembl/              # ChEMBL MCP server
│   │   ├── index.ts
│   │   ├── package.json
│   │   ├── tsconfig.json
│   │   └── Dockerfile
│   └── [future-server]/     # Additional MCP servers
├── package.json             # Root workspace config
├── pnpm-workspace.yaml      # pnpm workspace definition
└── tsconfig.json            # Base TypeScript config

Setup

1. Install pnpm (if not already installed):

npm install -g pnpm

2. Install dependencies:

pnpm install

3. Build all servers:

pnpm build

Or build a specific server:

pnpm build:chembl

Running Servers Locally

ChEMBL Server (stdio MCP)

pnpm --filter @mcp-mono/chembl start

Add to an MCP client (example):

{
  "mcpServers": {
    "chembl": {
      "command": "node",
      "args": ["/absolute/path/to/ChEMBL-MCP-Server/build/chembl/index.js"],
      "env": {}
    }
  }
}

ChEMBL Server (HTTP/SSE)

pnpm --filter @mcp-mono/chembl start:http

The server binds to 0.0.0.0:${PORT:-8080} and exposes HTTP/SSE endpoints.

Development

Watch mode for development:

# All servers
pnpm dev

# Specific server
pnpm dev:chembl

Deployment

Automatic Deployment (Dev)

Push to main branch triggers automatic deployment to Cloud Run for changed servers only.

The CI/CD pipeline:

1. Detects which servers have changes
2. Builds and pushes Docker images for changed servers
3. Deploys to Cloud Run as separate services (e.g., chembl-mcp-dev)

Manual Deployment (Production)

Use GitHub Actions workflow dispatch:

1. Go to Actions → "Build and Deploy MCP Servers (Production)"
2. Click "Run workflow"
3. Enter servers to deploy (e.g., chembl or all)

Each server deploys to its own Cloud Run service (e.g., chembl-mcp-prod)

Docker

Build a specific server:

docker build -f src/chembl/Dockerfile -t chembl-mcp-server .
docker run -p 8080:8080 chembl-mcp-server

Adding New MCP Servers

See ADDING_NEW_SERVER.md for a comprehensive step-by-step guide.

Quick summary:

1. Create src/your-server-name/ directory
2. Add required files: index.ts, package.json, tsconfig.json, Dockerfile
3. Update root package.json with build scripts
4. Test locally with Docker
5. Commit and push - CI/CD automatically deploys!

Critical requirements:

• Use @mcp-mono/your-server-name as package name
• Create node_modules symlink in Dockerfile: RUN ln -s /app/src/your-server-name/node_modules /app/build/your-server-name/node_modules
• Bind HTTP server to '0.0.0.0' for Cloud Run compatibility
• Parse PORT as integer: parseInt(process.env.PORT || '8080', 10)

Notes

• Each MCP server runs as a separate Cloud Run service for isolation and independent scaling
• Only changed servers are deployed automatically
• ChEMBL Data source: ChEMBL REST API (https://www.ebi.ac.uk/chembl/api/data)
• No API key required for public ChEMBL endpoints (subject to availability and rate limits)