n8n Integration

This guide shows you how to add FlautoPsy monitoring to your n8n automation workflows.

Step 1: Get Your Webhook URL

1. Log in to FlautoPsy at https://flautopsy.stralocroft.com/dashboard
2. Go to Settings → Setup (or /setup)
3. Copy your unique webhook URL

Step 2: Add an HTTP Request Node

1. Open your n8n workflow
2. Click + to add a new node
3. Search for HTTP Request and select it
4. Click Add node

Step 3: Configure the HTTP Request Node

In the HTTP Request node settings:

| Field | Value | |-------|-------| | Request Method | POST | | URL | Paste your FlautoPsy webhook URL | | Content-Type | application/json | | Headers | None required (n8n sets Content-Type automatically) |

Step 4: Set Request Body

Click on the Body tab and switch to JSON mode (if not already selected).

Paste the following JSON template:

{
  "workflow_id": "your-workflow-name",
  "prompt": "{{$node.PreviousNode.json.prompt}}",
  "output": "{{$node.PreviousNode.json.output}}",
  "latency_ms": {{$node.PreviousNode.json.latencyMs}},
  "cost_tokens": {{$node.PreviousNode.json.costTokens}},
  "model": "gpt-4"
}
 

Mapping n8n Variables

Replace the template variables with outputs from your LLM node:

| Field | n8n Variable | |-------|--------------| | workflow_id | A fixed string, e.g.,

"email-classifier"

{{$node.YourLLMNode.json.prompt}}

YourLLMNode

{{$node.YourLLMNode.json.message}}

{{$node.YourLLMNode.json.duration}}

{{add($node.YourLLMNode.json.usage.prompt_tokens, $node.YourLLMNode.json.usage.completion_tokens)}}

"gpt-4"

Example with OpenAI Node

If your workflow includes an OpenAI Chat Model node named

OpenAI

{
  "workflow_id": "email-classifier",
  "prompt": "{{$node.OpenAI.json.messages[0].content}}",
  "output": "{{$node.OpenAI.json.choices[0].message.content}}",
  "latency_ms": 245,
  "cost_tokens": {{add($node.OpenAI.json.usage.prompt_tokens, $node.OpenAI.json.usage.completion_tokens)}},
  "model": "gpt-4"
}
 

Step 5: Test the Node

1. Click Test step (play icon) to send a test request
2. You should see a Success response with status 200
3. Check FlautoPsy dashboard → Traces to confirm the trace arrived

Step 6: Activate the Workflow

1. Run your workflow 10+ times to build a baseline
2. Traces will be logged to FlautoPsy after each run
3. Once mature (≥10 traces), drift alerts will trigger automatically

n8n-Specific Notes

• Use
  text

  Copy

  {{}}

  for variable substitution, not
  text

  Copy

  ${}

  or other formats
• The HTTP Request node automatically sets
  text

  Copy

  Content-Type: application/json

• Test individual nodes by clicking Test step
• If your workflow has multiple branches, place the HTTP Request node after your LLM call

Common Issues

"Request failed: 401 Unauthorized"

• Double-check your webhook URL is correct
• Copy it again from FlautoPsy Settings → Setup
• Ensure no typos in the workspace_id parameter

Variables Not Showing in Output

• The LLM node must run before the HTTP Request node
• Check that your LLM node actually outputs the fields you're trying to access
• Use
  text

  Copy

  Test step

  on the LLM node first to verify output format

Latency Not Recorded

• n8n's execution time is available via
  text

  Copy

  $execution.runData

  , but it may vary
• You can hard-code a value (e.g.,
  text

  Copy

  250

  ) for testing
• Or calculate it manually:
  text

  Copy

  {{(new Date()).getTime() - $node.YourLLMNode.json.startTime}}

Troubleshooting

Workflow runs but traces don't appear

• Check that all 10 required fields are in the JSON body
• Verify the HTTP Request node shows a 200 OK response
• Check FlautoPsy dashboard → Traces tab (may take 5-10 seconds to appear)

How do I see what was sent?

• Use the Output tab in the HTTP Request node to inspect the response
• Or use n8n's Debug mode to see the full request/response

Next Steps

• View your workflow in FlautoPsy
• Learn about drift detection types
• Read FAQs for more questions