How to Connect OpenClaw AI Agent to Instagram DMs with InstantDM

OpenClaw is an open-source autonomous AI agent framework with over 200K GitHub stars. It connects to LLMs like Claude, GPT, or DeepSeek and executes multi-step tasks through messaging platforms. By connecting OpenClaw to InstantDM's webhook system, you can build a fully autonomous Instagram DM agent that qualifies leads, answers questions, books appointments, and manages customer conversations - all running locally on your machine with the LLM of your choice.

---

Why OpenClaw for Instagram DMs?

Advantage	Details
Any LLM backend	Use Claude, GPT-4o, DeepSeek, or local models via Ollama
Skills system	Modular skills for product lookup, CRM sync, appointment booking
Runs locally	Your data stays on your machine
Open source (MIT)	Full control, no vendor lock-in
Multi-platform	Same agent handles Telegram, WhatsApp, Discord, and now Instagram DMs
Persistent memory	Conversation history stored locally across sessions

---

Architecture

Instagram User sends DM
      |
      v
  InstantDM (receives via Meta Graph API)
      |
      v  Webhook POST
  Bridge Server (Node.js/Python)
      |
      v  Forwards to OpenClaw
  OpenClaw Agent (local)
      |
      v  Calls LLM (Claude/GPT/DeepSeek)
  LLM generates response
      |
      v
  OpenClaw returns reply to Bridge
      |
      v  InstantDM API
  InstantDM delivers reply as Instagram DM

The bridge server sits between InstantDM and OpenClaw, translating webhook events into OpenClaw messages and sending replies back through the InstantDM API.

---

What You Need

Requirement	Details
InstantDM account	Trendsetter plan or higher (API access)
InstantDM API key	Settings > API
OpenClaw installed	github.com/openclaw/openclaw
LLM API key	Anthropic (Claude), OpenAI (GPT), or DeepSeek
Node.js 18+	For the bridge server

---

Step 1: Install and Configure OpenClaw

# Clone OpenClaw
git clone https://github.com/openclaw/openclaw.git
cd openclaw

# Install dependencies
npm install

# Configure your LLM provider
cp .env.example .env

Edit .env with your LLM provider:

# For Claude (recommended)
ANTHROPIC_API_KEY=sk-ant-your-key-here
DEFAULT_MODEL=claude-sonnet-4-20250514

# Or for OpenAI
# OPENAI_API_KEY=sk-your-key-here
# DEFAULT_MODEL=gpt-4o

# Or for DeepSeek
# DEEPSEEK_API_KEY=your-key-here
# DEFAULT_MODEL=deepseek-chat

---

Step 2: Create an Instagram DM Skill

OpenClaw uses a skills system where each skill is a directory with a SKILL.md file. Create a skill for handling Instagram DMs:

mkdir -p skills/instagram-dm-agent

Create skills/instagram-dm-agent/SKILL.md:

# Instagram DM Agent

You are an Instagram DM customer support agent for [Your Brand].

## Instructions

- Respond to customer questions about products, pricing, shipping, and returns
- Keep responses under 280 characters (Instagram DM best practice)
- Be friendly and conversational, not corporate
- Never make up product information or pricing
- If you cannot help, offer to connect them with a human team member
- Do not use markdown formatting (Instagram does not render it)
- Qualify leads by understanding their needs and budget

## Product Information

[Add your product catalog, pricing, FAQs here]

## Lead Qualification

When a potential customer shows buying intent:
1. Understand what they are looking for
2. Ask about their timeline
3. Share relevant product links
4. Offer to book a consultation if appropriate

## Escalation Rules

Escalate to a human when:
- Customer is angry or frustrated
- Question requires account-specific information
- Customer explicitly asks for a human
- Topic involves refunds or disputes

---

Step 3: Build the Bridge Server

This Node.js server receives InstantDM webhooks and routes them to OpenClaw, then sends OpenClaw's responses back via the InstantDM API.

const express = require('express');
const crypto = require('crypto');
const { spawn } = require('child_process');

const app = express();
app.use(express.json());

const INSTANTDM_API_KEY = process.env.INSTANTDM_API_KEY;
const OPENCLAW_PATH = process.env.OPENCLAW_PATH || './openclaw';

// Track active OpenClaw conversations
const sessions = new Map();

function verifySignature(req) {
  const sig = req.headers['x-webhook-signature'];
  if (!sig) return false;
  const expected = crypto
    .createHmac('sha256', INSTANTDM_API_KEY)
    .update(JSON.stringify(req.body))
    .digest('hex');
  return crypto.timingSafeEqual(Buffer.from(sig), Buffer.from(expected));
}

app.post('/webhook/instantdm', async (req, res) => {
  res.status(200).json({ received: true });
  if (!verifySignature(req)) return;

  const { event, data } = req.body;
  if (event === 'dm_received') {
    await handleDM(data);
  }
});

async function handleDM(data) {
  const { instagram_user_id, username, message_text } = data;

  try {
    // Send message to OpenClaw and get response
    const reply = await sendToOpenClaw(instagram_user_id, message_text);

    // Send reply back via InstantDM
    await sendReply(instagram_user_id, reply);
  } catch (err) {
    console.error('OpenClaw error:', err);
    await sendReply(instagram_user_id,
      "Hey! Quick technical issue. A team member will follow up soon."
    );
  }
}

async function sendToOpenClaw(userId, message) {
  return new Promise((resolve, reject) => {
    // Use OpenClaw CLI to process the message
    const proc = spawn('node', [
      `${OPENCLAW_PATH}/index.js`,
      '--skill', 'instagram-dm-agent',
      '--session', `ig-${userId}`,
      '--message', message,
      '--format', 'text',
    ]);

    let output = '';
    let error = '';

    proc.stdout.on('data', (data) => { output += data.toString(); });
    proc.stderr.on('data', (data) => { error += data.toString(); });

    proc.on('close', (code) => {
      if (code !== 0) {
        reject(new Error(`OpenClaw exited with code ${code}: ${error}`));
      } else {
        resolve(output.trim());
      }
    });

    // Timeout after 30 seconds
    setTimeout(() => {
      proc.kill();
      reject(new Error('OpenClaw timeout'));
    }, 30000);
  });
}

async function sendReply(recipientId, text) {
  // Clean up for Instagram
  if (text.length > 1000) text = text.substring(0, 997) + '...';
  text = text.replace(/[*_~`#]/g, '');

  const res = await fetch('https://api.instantdm.com/api-webhook', {
    method: 'POST',
    headers: {
      'Content-Type': 'application/json',
      'Authorization': `Bearer ${INSTANTDM_API_KEY}`,
    },
    body: JSON.stringify({
      action: 'send_message',
      type: 'dm',
      recipient_id: recipientId,
      message: { type: 'text', text: text.trim() },
    }),
  });

  if (!res.ok) throw new Error(`InstantDM API error: ${res.status}`);
}

app.listen(3000, () => console.log('OpenClaw Instagram bridge running on port 3000'));

---

Step 4: Alternative - OpenClaw HTTP API Bridge

If you prefer using OpenClaw's HTTP API mode instead of the CLI, you can run OpenClaw as a server and communicate via HTTP:

const OPENCLAW_API = 'http://localhost:3001'; // OpenClaw API server

async function sendToOpenClaw(userId, message) {
  const response = await fetch(`${OPENCLAW_API}/api/chat`, {
    method: 'POST',
    headers: { 'Content-Type': 'application/json' },
    body: JSON.stringify({
      session_id: `ig-${userId}`,
      skill: 'instagram-dm-agent',
      message: message,
    }),
  });

  if (!response.ok) {
    throw new Error(`OpenClaw API error: ${response.status}`);
  }

  const data = await response.json();
  return data.reply;
}

Start OpenClaw in API mode:

cd openclaw
node index.js --serve --port 3001

---

Step 5: Add Custom Skills for Instagram Use Cases

Product Lookup Skill

Create skills/product-lookup/SKILL.md:

# Product Lookup

When a user asks about a product, search the catalog and return details.

## Tools

- search_products: Search by name, category, or keyword
- check_stock: Check availability for a specific product and size

## Response Format

Return product name, price, available sizes, and a direct link.
Keep the response under 280 characters.

Appointment Booking Skill

Create skills/book-appointment/SKILL.md:

# Appointment Booking

Help users book consultations or demos.

## Steps

1. Ask what they need help with
2. Suggest available time slots
3. Collect their name and email
4. Confirm the booking

## Tools

- get_available_slots: Returns open calendar slots
- create_booking: Books an appointment

---

Step 6: Connect OpenClaw to Your CRM

OpenClaw can sync lead data to your CRM as part of the conversation flow. Add a CRM skill:

# CRM Sync

After qualifying a lead, save their information to the CRM.

## Tools

- create_lead: Create a new lead in HubSpot/Salesforce
- update_lead: Update existing lead with new information
- tag_lead: Add tags based on conversation outcome (hot/warm/cold)

The bridge server can also fire webhooks to your CRM directly:

async function onLeadQualified(userId, leadData) {
  // Tag the contact in InstantDM
  await fetch('https://api.instantdm.com/api-webhook', {
    method: 'POST',
    headers: {
      'Content-Type': 'application/json',
      'Authorization': `Bearer ${INSTANTDM_API_KEY}`,
    },
    body: JSON.stringify({
      action: 'tag_contact',
      recipient_id: userId,
      tags: [leadData.score, leadData.interest],
    }),
  });

  // Send to your CRM webhook
  await fetch('https://your-crm-webhook.com/leads', {
    method: 'POST',
    headers: { 'Content-Type': 'application/json' },
    body: JSON.stringify({
      source: 'instagram_dm',
      user_id: userId,
      ...leadData,
    }),
  });
}

---

LLM Recommendations for OpenClaw + Instagram

Model	Best For	Cost
Claude Sonnet 4	Best overall quality for DM conversations	~$3/1K input, $15/1K output
GPT-4o-mini	High volume, cost-effective	~$0.15/1M input
DeepSeek V3	Budget option with good quality	~$0.14/1M input
Hermes 3 8B (Ollama)	Free, self-hosted, full privacy	$0 (local GPU)

---

Deployment

Run Locally (Development)

# Terminal 1: Start OpenClaw
cd openclaw && node index.js --serve --port 3001

# Terminal 2: Start the bridge server
INSTANTDM_API_KEY=your-key OPENCLAW_API=http://localhost:3001 node bridge.js

Deploy to Railway

# Push both OpenClaw and the bridge server to a repo
git push origin main

# Connect to Railway - it will detect the Node.js app
# Set environment variables in Railway dashboard:
# INSTANTDM_API_KEY, ANTHROPIC_API_KEY (or your LLM key)

Deploy to a VPS (DigitalOcean, Hetzner)

# Install Node.js and clone your repo
curl -fsSL https://deb.nodesource.com/setup_20.x | sudo bash -
sudo apt install -y nodejs
git clone https://github.com/your-repo/openclaw-instagram-bridge.git
cd openclaw-instagram-bridge
npm install

# Run with PM2 for process management
npm install -g pm2
pm2 start bridge.js --name openclaw-bridge
pm2 start openclaw/index.js --name openclaw-agent -- --serve --port 3001
pm2 save

---

Troubleshooting

Issue	Fix
OpenClaw not responding	Check if OpenClaw is running: curl http://localhost:3001/health
Slow responses	Use a faster LLM (GPT-4o-mini or Claude Haiku)
Skill not loading	Verify SKILL.md exists in the correct directory
Webhook not firing	Check InstantDM Settings > API for correct webhook URL
LLM rate limited	Add retry logic with exponential backoff in the bridge
Responses too long	Add character limit instructions to your SKILL.md

---

What's Next

• Install OpenClaw and create your Instagram DM skill
• Deploy the bridge server with a public URL
• Test with real Instagram DMs
• Add product lookup and appointment booking skills
• Connect to your CRM with HubSpot or Salesforce
• Set up Slack notifications for human handoff alerts