# 🌉 Bridge MCP Server
**"You don't need another automation tool. You just need a bridge."**
Connect Claude, Gemini, or any AI to your business tools via the Model Context Protocol (MCP).
---
## What is this?
Bridge is a hosted MCP server that gives AI direct access to your tools:
- **ClickUp** - Tasks, projects, time tracking, templates
- **Skills Library** - Pre-built instructions for common workflows
- **(Coming soon)** HubSpot, Guru, Google Analytics, and more
No workflows to build. No operations to count. Just ask your AI what you need.
---
## Architecture
```
┌─────────────────────────────────────────────────────────────────┐
│ │
│ Your AI (Claude/Gemini) │
│ │ │
│ ▼ │
│ ┌─────────────────┐ │
│ │ Bridge MCP │ ◄── Hosted on Google Cloud Run │
│ │ Server │ │
│ └────────┬────────┘ │
│ │ │
│ ┌──────┴──────┐ │
│ ▼ ▼ │
│ ClickUp Skills │
│ API Library │
│ │
└─────────────────────────────────────────────────────────────────┘
```
---
## For Claude Code: Deployment Instructions
Hey Claude Code! Here's how to deploy this server to Google Cloud Run.
### Prerequisites
1. Google Cloud CLI (`gcloud`) installed and authenticated
2. A Google Cloud project with billing enabled
3. Cloud Run and Cloud Build APIs enabled
### Step-by-Step Deployment
```bash
# 1. Navigate to this directory
cd /path/to/bridge-mcp
# 2. Make sure gcloud is authenticated
gcloud auth login
# 3. Set your project
gcloud config set project YOUR_PROJECT_ID
# 4. Enable required APIs
gcloud services enable cloudbuild.googleapis.com run.googleapis.com
# 5. Make deploy script executable and run it
chmod +x deploy.sh
./deploy.sh
```
### Or Deploy Manually
```bash
# Build the container
gcloud builds submit --tag gcr.io/bridge-mcp/bridge-mcp
# Deploy to Cloud Run
gcloud run deploy bridge-mcp \
--image gcr.io/bridge-mcp/bridge-mcp \
--platform managed \
--region us-central1 \
--allow-unauthenticated \
--memory 512Mi \
--set-env-vars "SUPABASE_URL=...,SUPABASE_KEY=..."
```
### After Deployment
You'll get a URL like: `https://bridge-mcp-xxxxx-uc.a.run.app`
- Health check: `https://YOUR_URL/health`
- MCP endpoint: `https://YOUR_URL/mcp`
---
## Connecting to Claude
Once deployed, add this MCP server to Claude:
1. Go to Claude settings → Integrations → MCP Servers
2. Add new server with URL: `https://YOUR_URL/mcp`
3. Name it "Bridge"
Now you can ask Claude things like:
- "Get my ClickUp task templates" (requires your ClickUp API key)
- "Create a task from my Project Kickoff template"
- "Load the transcript-to-tasks skill"
---
## Project Structure
```
bridge-mcp/
├── src/
│ ├── index.ts # Main server entry point
│ └── connectors/
│ ├── clickup.ts # ClickUp API tools
│ └── skills.ts # Skills library tools
├── package.json
├── tsconfig.json
├── Dockerfile
├── deploy.sh
└── README.md
```
---
## Adding New Connectors
To add a new connector (e.g., HubSpot):
1. Create `src/connectors/hubspot.ts`
2. Export a `registerHubSpotTools(server: McpServer)` function
3. Import and call it in `src/index.ts`
4. Rebuild and redeploy
---
## Local Development
```bash
# Install dependencies
npm install
# Build TypeScript
npm run build
# Run locally
npm start
# Server runs at http://localhost:8080
```
---
## Skills Library
Bridge can fetch skills (instruction files) from GitHub repos.
Default repo: `amara-ai/bridge-skills` (update in `src/connectors/skills.ts`)
Skills are markdown files that teach AI how to perform tasks:
- `transcript-to-tasks` - Turn meeting transcripts into ClickUp tasks
- `client-onboarding` - Standard client onboarding workflow
- `daily-standup-summary` - Summarize team progress
---
## Environment Variables
| Variable | Description | Default |
|----------|-------------|---------|
| `PORT` | Server port | `8080` |
Note: API keys (ClickUp, HubSpot, etc.) are passed per-request by users, not stored on the server.
---
## License
MIT - Built by AI Foundations by Amara
*"I prompt like shit. But I get glorious results."*
