n8n-nodes-anycast-agents

Connect n8n workflows to private infrastructure via Anycast Agents.

Query private databases, send messages to AI models behind firewalls, and invoke agent pools — all from your n8n workflows. No VPN, no port forwarding, no exposed ports.

Zapier connects 6,000 apps. Anycast connects everything else.

Install

In your n8n instance:

1. Go to Settings > Community Nodes
2. Enter n8n-nodes-anycast-agents
3. Click Install

Or via CLI:

cd ~/.n8n
npm install n8n-nodes-anycast-agents

Credential Setup

1. Sign up at agents.anycast.com
2. Go to Settings > API Keys and create a new key
3. In n8n, go to Credentials > New Credential > Anycast Agents API
4. Paste your API key

Nodes

Anycast Agent

Main node with three resources:

Agent operations:

Token operations:

Group operations:

Anycast Agent Trigger

Polling trigger node:

Filter by specific agent ID or group. Configurable polling: 10s, 30s, 1m, 5m.

Key Use Cases

1. Query a Private Database from n8n

Your Postgres is behind a firewall. An Anycast agent runs next to it.
n8n queries it through the agent — no port forwarding needed.

[Schedule Trigger] → [Anycast Agent: Query Connector]
                       connectorId: "prod-postgres"
                       query: "SELECT * FROM orders WHERE status = 'pending'"
                   → [Each row becomes an n8n item]
                   → [Send Email: daily report]

2. Route Questions to Your Private AI

A Claude or GPT instance runs on your own hardware.
n8n routes user questions through an Anycast agent.

[n8n Form Trigger] → [Anycast Agent: Send Message]
    "Ask a question"     agentId: "my-claude-agent"
                         message: { "prompt": "{{$json.question}}" }
                     → [Wait for AI response (30s timeout)]
                     → [Slack: Send Response]

The agent runs the anycast-agent Python SDK:

from anycast_agent import AnycastAgent
import anthropic

agent = AnycastAgent(token="agt_xxx", rendezvous_url="wss://...")
client = anthropic.Anthropic()

@agent.on_message
async def handle(from_agent, data):
    response = client.messages.create(
        model="claude-sonnet-4-20250514",
        messages=[{"role": "user", "content": data["prompt"]}]
    )
    await agent.send(from_agent, {"answer": response.content[0].text})

await agent.connect()
await agent.run_forever()

3. Fan-Out to Multiple AI Models

Ask the same question to multiple models, aggregate responses:

[Webhook Trigger] → [Anycast Agent: Invoke Pool]
                       groupId: "ai-models"
                       message: { "prompt": "..." }
                       waitForAll: true
                   → [Each model's response is a separate item]
                   → [Code: Compare and select best answer]
                   → [HTTP Response: Return to caller]

4. Monitor Private Infrastructure

Agents report health from behind firewalls:

[Anycast Agent Trigger]     → [IF: Status is "offline"]
  event: "Status Changed"   → [Slack: Alert #ops channel]
  groupId: "production"     → [PagerDuty: Create incident]

5. Scheduled Private Data Sync

Pull data from internal systems on a schedule:

[Schedule: Every hour] → [Anycast Agent: Query Connector]
                            query: "SELECT * FROM inventory WHERE updated_at > NOW() - INTERVAL '1 hour'"
                       → [Airtable: Update rows]
                       → [Slack: "Inventory synced: {{$json.rowCount}} rows"]

How It Works

1. Install the any binary or SDK on your private machine
2. It connects outbound to the Anycast rendezvous network (no inbound ports)
3. n8n sends requests through the Anycast API
4. The agent receives the request, executes it locally, returns the response
5. n8n gets structured data back — each row/response becomes an n8n item

All traffic is encrypted end-to-end. Anycast never sees your data.

API Reference

• Agents API
• Connectors API
• OpenAPI Spec

License

MIT