Antigravity MCP Bridge
README.md
<div align="center">
# ⚡ Gemini Antigravity Bridge
### The World's First Bidirectional MCP Bridge Between Google Gemini Spark & DeepMind Antigravity
[](https://pypi.org/project/gemini-antigravity-bridge/)
[](https://doi.org/10.5281/zenodo.22259022)
[](https://glama.ai/mcp/servers/nandhakumar-murugan/gemini-antigravity-bridge)
[](https://github.com/punkpeye/awesome-mcp-servers#coding-agents)
[](https://python.org)
[](https://modelcontextprotocol.io)
[](LICENSE)
[](https://discuss.ai.google.dev/t/show-tell-gemini-antigravity-bridge-bidirectional-mcp-agent-orchestration-between-gemini-spark-deepmind-antigravity/180640/1)
[](tests/)
[](https://gemini.google.com)
[](https://deepmind.google)
[](https://ai.google.dev/)
[](https://github.com/nandhakumar-murugan/gemini-antigravity-bridge/stargazers)
[](https://github.com/nandhakumar-murugan/gemini-antigravity-bridge/discussions)
**Connect Google's cloud AI (Gemini Spark powered by Gemini 3.8 / 3.7) to your local agentic IDE (Antigravity) with 23 powerful MCP tools — enabling autonomous task dispatch, shared memory, file operations, and cross-agent orchestration.**
[Quick Start](#-quick-start) · [Architecture](#-architecture) · [Tools](#-available-tools-23) · [Connected Apps](#-spark-connected-apps) · [Deploy](#-deployment)
</div>
---
## 🌐 What Is This?
**Gemini Antigravity Bridge** is an MCP (Model Context Protocol) server that creates a **persistent, bidirectional communication channel** between:
- ☁️ **Google Gemini Spark** (cloud-based AI assistant with Google Workspace access, powered by Gemini 3.8 / 3.7)
- 💻 **Google DeepMind Antigravity** (local agentic IDE with full system access)
Optimized for **Gemini 3.8's next-generation agentic reasoning**, this enables a **fully autonomous loop** where Spark can dispatch coding tasks to your local machine, and Antigravity can request cloud intelligence back from Spark — all through a standardized, secure protocol.
```
┌──────────────────────────────────────────────────────────────────┐
│ GEMINI SPARK (Google Cloud) │
│ 📬 Gmail │ 📁 Drive │ 📅 Calendar │ 🎨 Canva │ 🎥 YouTube │
└─────────────────────────┬────────────────────────────────────────┘
│ MCP over HTTPS
▼
┌──────────────────────────────────────────────────────────────────┐
│ ⚡ GEMINI ANTIGRAVITY BRIDGE (MCP Server) │
│ │
│ 🔧 System Commands 📂 File Operations 🤖 Agent Dispatch │
│ 🧠 Shared Memory 🔄 Project Sync 📊 Status Reports │
│ │
│ Tunnel: ngrok / Cloudflare (permanent HTTPS domain) │
└─────────────────────────┬────────────────────────────────────────┘
│ Local MCP + Message Injection
▼
┌──────────────────────────────────────────────────────────────────┐
│ ANTIGRAVITY IDE (Local Machine) │
│ 🖥️ Terminal │ 📝 Code Editor │ 🌳 Git │ 🧪 Tests │ 🚀 Deploy │
└──────────────────────────────────────────────────────────────────┘
```
---
## 🏆 Why This Bridge?
| Feature | This Project | Other MCP Servers |
|---|:---:|:---:|
| Bidirectional Cloud ↔ Local | ✅ | ❌ |
| Autonomous Agent Task Dispatch | ✅ | ❌ |
| Cross-Agent Shared Memory | ✅ | ❌ |
| Antigravity Conversation Injection | ✅ | ❌ |
| Spark Connected Apps Orchestration | ✅ | ❌ |
| Google Workspace Integration | ✅ (via Spark) | Partial |
| Local File System Access | ✅ | Some |
| System Command Execution | ✅ | Some |
---
## 🚀 Quick Start
### Prerequisites
- Python 3.10+
- [ngrok account](https://ngrok.com) (free tier) with a reserved domain
- Google Gemini Spark (Advanced/Ultra subscription)
- Google DeepMind Antigravity IDE
### Installation
```bash
# Clone the repository
git clone https://github.com/nandhakumar-murugan/gemini-antigravity-bridge.git
cd gemini-antigravity-bridge
# Install dependencies or install package directly
pip install -r requirements.txt
# OR install as a CLI tool:
pip install -e .
# Configure environment
cp .env.example .env
# Edit .env with your NGROK_AUTHTOKEN and NGROK_DOMAIN
# Launch the bridge (via script or CLI)
gemini-bridge
# or: python run_with_tunnel.py
```
### Connect to Gemini Spark
1. Open [Gemini Spark](https://gemini.google.com) → Settings → Connected Apps
2. Click **"Add a custom app"**
3. Enter your public MCP URL: `https://your-domain.ngrok-free.dev/mcp`
4. Spark will discover all 23 tools automatically!
---
## 🛠️ Available Tools (23)
### 🔧 System Execution
| Tool | Description |
|---|---|
| `run_system_command` | Execute any shell/PowerShell command with captured stdout/stderr |
| `run_batch_commands` | Run multiple commands sequentially with error handling |
### 📂 File Operations
| Tool | Description |
|---|---|
| `read_file` | Read file contents from any path on the local machine |
| `write_file` | Create or overwrite files with specified content |
| `edit_file` | Surgically edit specific lines in existing files |
| `append_file` | Append content to the end of a file |
| `batch_write_files` | Create multiple files in a single operation |
| `create_full_project` | Scaffold an entire project directory structure |
| `list_directory` | List directory contents with metadata |
### 🤖 Agent Orchestration
| Tool | Description |
|---|---|
| `run_agent_task` | Launch an autonomous Antigravity coding subagent |
| `get_agent_status` | Check status and output of a running agent task |
| `terminate_task` | Kill a running agent task |
### 🧠 Shared Memory & Sync
| Tool | Description |
|---|---|
| `get_bridge_history` | Retrieve full cross-client operation history |
| `save_session_note` | Write a structured note to shared memory (tagged) |
| `get_session_notes` | Read session notes by tag or date range |
| `sync_project_to_gemini` | Sync project metadata to Spark's knowledge base |
| `git_quick_status` | Quick Git status check for any repository |
### 🌐 Spark Connected Apps
| Tool | Description |
|---|---|
| `request_spark_connected_app_action` | Dispatch tasks to Spark's connected apps (@Canva, @YouTube, @Gmail, etc.) |
| `get_spark_connected_apps_catalog` | List all available connected apps and capabilities |
### 🔗 Antigravity Deep Integration
| Tool | Description |
|---|---|
| `list_antigravity_conversations` | List all active Antigravity IDE conversations |
| `inject_message` | Inject a structured message directly into an Antigravity conversation |
| `send_spark_to_antigravity_task` | Dispatch a full task brief from Spark to Antigravity |
| `get_antigravity_agent_report` | Get the latest execution report for Spark to review |
---
## 🔄 Spark Connected Apps
Through the bridge, Antigravity can orchestrate Spark's connected Google & third-party apps:
| App | Capabilities |
|---|---|
| 🎨 **@Canva** | Poster design, infographics, slide decks |
| 📁 **@Google Drive** | Cloud file search, folder management |
| 📝 **@Google Docs** | Document creation and collaboration |
| 📌 **@Google Keep** | Quick notes, flashcards, checklists |
| 🎥 **@YouTube** | Video search, transcript extraction |
| 📬 **@Gmail** | Email reading, URL extraction |
| 📓 **@Gemini Notebook** | Deep research and synthesis |
| 📦 **@Dropbox** | Cloud storage sync via MCP |
---
## 🏗️ Architecture
```mermaid
graph TB
subgraph Cloud["☁️ Google Cloud"]
Spark["Gemini Spark<br/>Consumer AI Assistant"]
Gmail["📬 Gmail"]
Drive["📁 Google Drive"]
Calendar["📅 Google Calendar"]
Canva["🎨 Canva"]
end
subgraph Bridge["⚡ Antigravity Bridge"]
MCP["MCP Server<br/>(FastMCP + Starlette)"]
Tunnel["HTTPS Tunnel<br/>(ngrok / Cloudflare)"]
Memory["🧠 Shared Memory<br/>(bridge_history.json)"]
Dashboard["📊 Web Dashboard"]
end
subgraph Local["💻 Local Machine"]
AGY["Antigravity IDE"]
Terminal["🖥️ Terminal"]
Git["🌳 Git Repos"]
Files["📂 File System"]
end
Spark -->|"MCP over HTTPS"| Tunnel
Tunnel --> MCP
MCP -->|"inject_message"| AGY
MCP -->|"run_system_command"| Terminal
MCP -->|"git_quick_status"| Git
MCP -->|"read/write_file"| Files
MCP --> Memory
AGY -->|"save_session_note"| Memory
Memory -->|"get_agent_report"| Spark
Spark --> Gmail
Spark --> Drive
Spark --> Calendar
Spark --> Canva
```
---
## 🚢 Deployment
### Auto-Start on Windows Boot
The bridge includes a silent startup script that launches automatically when you log in:
```
📁 %APPDATA%\Microsoft\Windows\Start Menu\Programs\Startup\
└── GeminiAntigravityBridge.vbs (silent background launcher)
```
### Cloud Deployment (24/7 Uptime)
```bash
# Using Docker
docker build -t gemini-antigravity-bridge .
docker run -d -p 8000:8000 --env-file .env gemini-antigravity-bridge
# Using Render.com / Google Cloud Run
# Push to GitHub → Connect repo → Deploy automatically
```
---
## 📋 Scheduled Monitoring
The bridge supports multi-tier automated monitoring via Gemini Spark schedules:
| Tier | Cadence | Purpose |
|---|---|---|
| 🔴 **Instant Trigger** | Real-time (on email arrival) | Zero-delay alerts for college & NPTEL emails |
| 🟡 **Hourly Sweep** | Every 60 minutes | Broad monitoring across all platforms |
| 🟢 **Morning Briefing** | Daily @ 8:00 AM | Strategic daily plan with exam schedules |
---
## 🤝 Contributing
Contributions are welcome! Please read the [Apache 2.0 License](LICENSE) before submitting PRs.
1. Fork the repository
2. Create your feature branch (`git checkout -b feat/amazing-feature`)
3. Commit your changes (`git commit -m 'feat: add amazing feature'`)
4. Push to the branch (`git push origin feat/amazing-feature`)
5. Open a Pull Request
---
## 📄 License
This project is licensed under the **Apache License 2.0** — see the [LICENSE](LICENSE) file for details.
---
## 📚 Research Paper & Academic Citation
This framework implements the formal architecture introduced in the published research paper:
> **AgentShield: A Zero-Trust Runtime Guardrail Architecture for Autonomous Multi-Agent AI Systems with Bidirectional Context Synchronization**
> *Nandhakumar Murugan*
> Department of Computer Science and Engineering (Cybersecurity), KGiSL Institute of Technology (KiTE), Autonomous
> **Permanent DOI**: [10.5281/zenodo.22259022](https://doi.org/10.5281/zenodo.22259022) | **Open Access**: [Zenodo CERN Repository](https://zenodo.org/records/22259022)
If you reference or build upon this architecture in academic or enterprise research, please cite:
```bibtex
@article{murugan2026agentshield,
title={AgentShield: A Zero-Trust Runtime Guardrail Architecture for Autonomous Multi-Agent AI Systems with Bidirectional Context Synchronization},
author={Murugan, Nandhakumar},
journal={Zenodo},
year={2026},
doi={10.5281/zenodo.22259022},
url={https://doi.org/10.5281/zenodo.22259022}
}
```
---
## 👨💻 Author
**Nandhakumar Murugan**
B.E. Computer Science & Cyber Security | KGiSL Institute of Technology
Google Student Ambassador (GID: 36) | Open Source Contributor
[](https://github.com/nandhakumar-murugan)
[](https://linkedin.com/in/nandhakumar-murugan)
---
<div align="center">
**Built with ❤️ using Google Gemini, DeepMind Antigravity & Model Context Protocol**
</div>
TDQS
C2.9/5.0
Scored across 23 tools
Disambiguation2/5
Multiple tools have overlapping purposes, such as run_system_command vs run_batch_commands, write_file vs batch_write_files, and run_agent_task vs send_spark_to_antigravity_task vs inject_message. The boundaries between session notes, bridge history, and agent status reports are also unclear, creating significant potential for misselection.
Naming Consistency3/5
Most tool names follow a verb_noun pattern, but the verbs are inconsistent across similar actions (e.g., run_agent_task, send_spark_to_antigravity_task, inject_message) and some names like git_quick_status do not follow the pattern. The naming is readable but not systematically consistent.
Tool Count3/5
With 23 tools, the set is at the high end of the 'heavy but not extreme' range. The broad scope of filesystem, shell, agent, notes, and external app operations partially justifies the count, but it feels larger than necessary and could be consolidated.
Completeness4/5
The toolset covers core filesystem operations, command execution, batch operations, agent task management, session notes, and external app integration. Minor gaps exist such as no explicit delete file/note tool and no direct list-agent-tasks function, but the primary workflows are well represented.
Maintenance
ActivityMaintained
ResponsivenessNo issues