Skip to content
Documentation
Connect a clientn8n

n8n

Configure Guard API requests and automation in n8n.

Set up with your own AI
Model protocol

n8n is a (no-code) automation platform. Claude via Guard connects in two ways: as an Anthropic credential for nodes inside a workflow, and as the model behind n8n's own AI Assistant.

Workflow node

  1. Create a Credential of type Anthropic API.
  2. API Key — your gd-… from the bot.
  3. Base URL — https://api.guardrelay.ai.
  4. In the workflow add an AI Agent or Anthropic Chat Model node and select this credential.
  5. Model — claude-sonnet-5-5 or claude-opus-5.

Built-in AI Assistant

This is not a node but n8n's own helper: it lives in the editor side panel, reads your workflows and edits them. It is configured with environment variables, not a credential. Requires self-hosted n8n 2.31.0 or newer.

yaml
services:
  n8n:
    image: docker.n8n.io/n8nio/n8n:latest
    environment:
      - N8N_ENABLED_MODULES=instance-ai
      - N8N_INSTANCE_AI_MODEL=anthropic/claude-sonnet-5-5
      - N8N_INSTANCE_AI_MODEL_URL=https://api.guardrelay.ai
      - N8N_INSTANCE_AI_MODEL_API_KEY=gd-…
bash
docker compose pull
docker compose up -d

Kimi, Composer and Grok

Kimi works both with an Anthropic credential (Base URL https://api.guardrelay.ai) and with a node on an OpenAI-compatible API (Base URL https://api.guardrelay.ai/v1). The key is the same gd-…, and the model field takes one of the four Kimi IDs. On the Anthropic address the accepted field list is closed: a stray field in the node settings gives a 400 before billing.

Composer and Grok accept both the Anthropic address https://api.guardrelay.ai and the OpenAI-compatible https://api.guardrelay.ai/v1. Put the exact id in the model field, for example cursor-grok-4.6-medium or composer-2.5.

A node for Kimi

  1. Create a Credential with an OpenAI-compatible API.
  2. API Key — your gd-… from the bot.
  3. Base URL — https://api.guardrelay.ai/v1, exactly with /v1.
  4. In the workflow add an AI Agent or OpenAI Chat Model node and select this credential.
  5. Model — type the id by hand: k3, k3-256k, kimi-for-coding or kimi-for-coding-highspeed.

Before configuring the node, check the key with one command:

bash
curl -X POST https://api.guardrelay.ai/v1/chat/completions \
  -H "Authorization: Bearer $GUARD_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "k3", "messages": [{"role": "user", "content": "Hello!"}], "max_tokens": 128}'

A node for Composer and Grok

  1. Create a Credential of type Anthropic API: both families accept the Anthropic Messages protocol.
  2. API Key — your gd-… from the bot.
  3. Base URL — https://api.guardrelay.ai, without /v1.
  4. Add an AI Agent or Anthropic Chat Model node and select this credential.
  5. Model — type the exact id: cursor-grok-4.6-high, cursor-grok-4.6-medium, composer-2.5 or their -fast variant.
bash
curl -X POST https://api.guardrelay.ai/v1/chat/completions \
  -H "Authorization: Bearer gd-your-key" \
  -H "Content-Type: application/json" \
  -d '{"model": "composer-2.5", "messages": [{"role": "user", "content": "Hi!"}], "max_tokens": 128}'

GLM in n8n through HTTP Request

This example sends a regular GLM HTTP request through an HTTP Request node. Specialized n8n AI nodes may require different settings.

  1. Create a credential/header secret for Authorization with value Bearer gd-….
  2. Method: POST; URL: https://api.guardrelay.ai/glm/v1/chat/completions.
  3. Add content-type: application/json.
  4. Choose a JSON body and send an exact model ID, messages, and max_tokens: 4096.
  5. Use choices[0].message.content in the next node.
json
{
  "model": "glm-5.3",
  "messages": [
    {"role": "user", "content": "{{$json.prompt}}"}
  ],
  "max_tokens": 4096
}

For stateful Responses, use another HTTP Request to /glm/v1/responses: save response.id, then pass it as previous_response_id. store defaults to true; continuation is unavailable after store: false.

Sections

On this page