lindy-sdk-patterns

Lindy SDK & Integration Patterns

Overview

Lindy is primarily a no-code platform. External integration happens through three channels: Webhook triggers (inbound), HTTP Request actions (outbound), and Run Code actions (inline Python/JS execution via E2B sandbox). This skill covers patterns for each.

Prerequisites

• Lindy account with active agents
• Node.js 18+ or Python 3.10+ for webhook receivers
• Completed lindy-install-auth setup

Pattern 1: Webhook Trigger Integration

Your application fires webhooks to wake Lindy agents:

// lindy-client.ts — Reusable Lindy webhook trigger client
class LindyClient {
  private webhookUrl: string;
  private secret: string;
 
  constructor(webhookUrl: string, secret: string) {
    this.webhookUrl = webhookUrl;
    this.secret = secret;
  }
 
  async trigger(payload: Record<

Pattern 2: HTTP Request Action (Agent Calling Your API)

Configure a Lindy agent to call your API as an action step:

In Lindy Dashboard — Add HTTP Request action:

• Method: POST
• URL: https://api.yourapp.com/process
• Headers: Authorization: Bearer {{your_api_key}}, Content-Type: application/json
• Body (AI Prompt mode):

  Send the processed data as JSON with fields matching the API schema.
  Include: name from {{trigger.data.name}}, analysis from previous step.
  

Your API endpoint receives the call:

// Your API receiving Lindy agent calls
app.post('/process', async (req, res) => {
  const { name, analysis } = req.body;
  const result = await processData(name, analysis);
  res.json({ result, processedAt: new Date().toISOString() });

Pattern 3: Run Code Action (E2B Sandbox)

Execute Python or JavaScript directly in Lindy workflows. Code runs in isolated Firecracker microVMs with ~150ms startup time.

Python example (data transformation in a workflow):

# Run Code action — Python
# Input variables: raw_data (string from previous step)
import json
 
data = json.loads(raw_data)  # Input vars are always strings
 
# Process
cleaned = [
    {"name": item["name"].strip(), "score": float(item["score"])}
    for item in data["items"

JavaScript example (API call + processing):

// Run Code action — JavaScript
// Input variables: query (string), api_key (string)
const response = await fetch(`https://api.example.com/search?q=${query}`, {
  headers: { 'Authorization': `Bearer ${api_key}` }
});
const data = await response.json();
 
const summary = data.results.

Run Code outputs (available to subsequent steps):

Available Python libraries: pandas, numpy, scipy, scikit-learn, matplotlib, requests, aiohttp, beautifulsoup4, nltk, spacy, openpyxl, python-docx

Key constraint: All input variables arrive as strings. Cast explicitly: count = int(count_str), data = json.loads(json_str)

Pattern 4: Callback Pattern (Async Two-Way)

Send a callbackUrl in your webhook payload. Lindy can respond back using the Send POST Request to Callback action:

// Your app triggers Lindy with a callback URL
await lindy.trigger({
  event: 'analyze.request',
  data: { text: 'Analyze this quarterly report...' },
  callbackUrl: 'https://api.yourapp.com/lindy-callback'
});
 
// Your callback handler receives Lindy's response
app.post('/lindy-callback', (req, res) => {
  const { analysis

Pattern 5: Retry with Exponential Backoff

async function triggerWithRetry(
  client: LindyClient,
  payload: Record<string, unknown>,
  maxRetries = 3
): Promise<void> {
  for (let attempt = 0; attempt 

Error Handling

Resources

• Calling Any API
• Run Code Documentation
• Webhooks Documentation

Next Steps

Proceed to lindy-core-workflow-a for full agent creation workflows.