Custom Integration

If you're using a platform not listed in our integration guides, you can integrate FlautoPsy with any application that can make HTTP requests.

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 (looks like
   text

   Copy

   https://flautopsy.stralocroft.com/api/webhook?workspace_id=...

   )

HTTP Request Format

Send a POST request to your webhook URL with the following JSON body:

URL

POST https://flautopsy.stralocroft.com/api/webhook?workspace_id=YOUR_WORKSPACE_ID
 

Headers

Content-Type: application/json
 

No authentication header is needed — your workspace ID is in the URL.

Request Body

{
  "workflow_id": "string (required)",
  "prompt": "string (optional)",
  "output": "string (required)",
  "latency_ms": number (required),
  "cost_tokens": number (required),
  "model": "string (required)"
}
 

Field Definitions

| Field | Type | Required | Description | |-------|------|----------|-------------| |

workflow_id

"email-classifier"

prompt

output

latency_ms

cost_tokens

model

"gpt-4"

"claude-3-opus"

Example: cURL

curl -X POST https://flautopsy.stralocroft.com/api/webhook?workspace_id=YOUR_WORKSPACE_ID \
  -H "Content-Type: application/json" \
  -d '{
    "workflow_id": "email-classifier",
    "prompt": "Classify this email as spam or not",
    "output": "not-spam",
    "latency_ms": 245,
    "cost_tokens": 1250,
    "model": "gpt-4"
  }'
 

Replace

YOUR_WORKSPACE_ID

Example: Python

import requests
import time
 
# Replace with your actual webhook URL
webhook_url = "https://flautopsy.stralocroft.com/api/webhook?workspace_id=YOUR_WORKSPACE_ID"
 
# Your LLM call happens here
start_time = time.time()
output = "not-spam"  # Your LLM result
latency_ms = int((time.time() - start_time) * 1000)
 
# Send to FlautoPsy
payload = {
    "workflow_id": "email-classifier",
    "prompt": "Classify this email as spam or not",
    "output": output,
    "latency_ms": latency_ms,
    "cost_tokens": 1250,
    "model": "gpt-4"
}
 
response = requests.post(webhook_url, json=payload)
print(response.status_code)  # Should be 200
 

Example: Node.js

const fetch = require('node-fetch');
 
async function sendToFlautoPsy() {
  const webhookUrl = 'https://flautopsy.stralocroft.com/api/webhook?workspace_id=YOUR_WORKSPACE_ID';
  
  const payload = {
    workflow_id: 'email-classifier',
    prompt: 'Classify this email as spam or not',
    output: 'not-spam',
    latency_ms: 245,
    cost_tokens: 1250,
    model: 'gpt-4'
  };
  
  const response = await fetch(webhookUrl, {
    method: 'POST',
    headers: { 'Content-Type': 'application/json' },
    body: JSON.stringify(payload)
  });
  
  console.log(response.status); // Should be 200
}
 

Response Format

On success, you'll receive a 200 OK response:

{
  "success": true,
  "trace_id": "trace_abc123def456",
  "message": "Trace recorded"
}
 

Error Responses

• 400 Bad Request — Missing required fields or invalid JSON
• 401 Unauthorized — Invalid workspace ID
• 429 Too Many Requests — Rate limit exceeded (contact support for limits)
• 500 Server Error — FlautoPsy internal error (rare)

Best Practices

1. Send traces as soon as possible after your LLM call completes
2. Use a consistent
   text

   Copy

   workflow_id

   for the same workflow (don't change it)
3. Don't retry failed requests indefinitely — if a request fails after 3 attempts, log it and move on
4. Measure latency accurately — use server-side timestamps, not client-side
5. Truncate long outputs to avoid oversized payloads (max 10,000 chars recommended)

Monitoring Traces

Once you've sent at least 10 traces:

1. Go to FlautoPsy dashboard → Workflows
2. Click your workflow to see:
   • Individual traces (Traces tab)
   • Alerts triggered (Alerts tab)
   • Baseline stats (Overview tab)

Troubleshooting

"405 Method Not Allowed"

• Ensure you're using POST, not GET or PUT

"400 Bad Request"

• Check that the JSON is valid (use a JSON validator)
• Verify all required fields are present
• Ensure
  text

  Copy

  latency_ms

  and
  text

  Copy

  cost_tokens

  are numbers

Traces not appearing

• Confirm the HTTP response status is 200
• Check that
  text

  Copy

  workflow_id

  is consistent across requests
• Wait 5-10 seconds and refresh the dashboard

Rate limit exceeded

• Contact support@stralocroft.com to discuss your usage pattern
• Consider batching multiple traces into a single request (max 10 per batch)

Next Steps

• View your traces
• Learn about baselines
• Check the API reference for more endpoints