n8n-nodes-agy

Official n8n Community Node to execute Google Antigravity CLI (AGY) AI agents directly inside your n8n workflows.

Transform your n8n instance into an autonomous AI agent engine—powering code generation, project refactoring, and multi-turn AI reasoning tasks seamlessly.

🚀 Quick Start in n8n

1. Install Node in n8n

1. Go to Settings ➡️ Community Nodes in your n8n dashboard.
2. Click Install a community node.
3. Enter n8n-nodes-agy and click Install.

2. Add AGY Node to Canvas

1. Search for AGY in your n8n workflow node selector.
2. Enter your prompt in the Prompt / Instruction field:

   "Analyze this JSON payload and draft a refactored Python backend function."

3. Click Test step — receive structured AI response and token metrics instantly!

🔑 AGY Authentication & .gemini Directory Setup

To run AGY inside n8n headlessly without interactive login prompts, the n8n Docker container accesses your host's AGY authentication session stored in ~/.gemini.

1. Initial Host Authentication

Before running the node in n8n, perform a one-time login on your host machine:

# Run AGY CLI on host to authenticate with Google
agy

This generates OAuth tokens (antigravity-oauth-token) and configuration files inside your host directory ~/.gemini.

2. Set Host Folder Permissions (~/.gemini)

Grant group read and write permissions on ~/.gemini so the n8n container process (running with your host GID via group_add) can read and refresh authentication tokens:

# Grant group read and write permissions to ~/.gemini
sudo chmod -R g+rw ~/.gemini

📥 Output Payload Schema (n8n Data Mapping)

The AGY node returns structured JSON data ready for downstream n8n nodes (HTTP Request, Slack, Email, Postgres, Webhook):

{
  "success": true,
  "exitCode": 0,
  "parsed": {
    "conversation_id": "8a32a4f7-5b5a-4631-9c24-46da12e853a3",
    "status": "SUCCESS",
    "response": "Here is the refactored Python function with optimized memory usage...",
    "duration_seconds": 3.094,
    "num_turns": 1,
    "usage": {
      "input_tokens": 9946,
      "output_tokens": 254,
      "thinking_tokens": 175,
      "cache_read_tokens": 8141,
      "total_tokens": 10200
    }
  },
  "prompt": "Refactor python function",
  "model": "Gemini 3.8 Flash (High)",
  "command": "agy --dangerously-skip-permissions --output-format json --model \"Gemini 3.8 Flash (High)\" -p \"Refactor python function\""
}

Accessing Output in n8n Expressions:

• AI Response Text: {{ $json.parsed.response }}
• Total Token Count: {{ $json.parsed.usage.total_tokens }}
• Conversation Session ID: {{ $json.parsed.conversation_id }}

⚡ Workflow Integration Examples

Example 1: Webhook ➡️ AGY AI ➡️ Slack Notification

1. Webhook Trigger: Receives code review request or user prompt via HTTP POST.
2. AGY Node: Runs prompt with Gemini 3.8 Flash (High) in headless unattended mode.
3. Slack Node: Sends {{ $json.parsed.response }} directly to your dev channel.

Example 2: Scheduled Daily Code Health Check

1. Schedule Trigger: Runs daily at 09:00 AM.
2. AGY Node: Executes system refactoring analysis on target repository.
3. Email / Notion Node: Archives daily AI code recommendations.

⚙️ n8n Node Options & Parameters

🐳 Docker Deployment Template for n8n Servers

If hosting n8n in Docker (Oracle Cloud, AWS, GCP, VPS), use this docker-compose.yml to mount the host agy binary, .gemini settings, and glibc libraries:

services:
  n8n:
    image: n8nio/n8n:latest
    container_name: n8n
    group_add:
      - "1001"                              # Grants host user group (e.g. 1001/ubuntu) access to .gemini files
    restart: unless-stopped
    ports:
      - "5678:5678"
    environment:
      - HOME=/home/node                     # Container HOME path
      - N8N_HOST=0.0.0.0
      - N8N_PORT=5678
      - N8N_PROTOCOL=http
      - NODE_ENV=production
      - WEBHOOK_URL=http://<YOUR_SERVER_IP>:5678
      - N8N_SECURE_COOKIE=false             # Allow HTTP IP login
    volumes:
      - n8n_data:/home/node/.n8n
      # Mount host AGY binary & authentication
      - ~/.local/bin/agy:/usr/local/bin/agy:ro
      - ~/.gemini:/home/node/.gemini
      # Mount host glibc dynamic libraries
      - /lib/aarch64-linux-gnu:/lib/aarch64-linux-gnu:ro
      - /lib/ld-linux-aarch64.so.1:/lib/ld-linux-aarch64.so.1:ro

volumes:
  n8n_data:

❓ Troubleshooting FAQs

• open /home/node/.gemini/antigravity-cli/...: permission denied: Ensure group_add: ["1001"] (or your host GID) is set in your docker-compose.yml and run sudo chmod -R g+rw ~/.gemini on the host.
• authentication required: Perform initial interactive login on host by running agy once before starting n8n.
• Error relocating /usr/local/bin/agy: symbol not found: Ensure host glibc libraries (/lib/aarch64-linux-gnu) are mounted in your docker-compose.yml.

📄 License

MIT License