AI App Maker Integrations

Goal: Connect no-code AI platforms to your governed data.

---

Supported Platforms

Platform	MCP Support	Integration
Dify	Native	MCP tool provider
Langflow	Via custom node	MCP client node
Flowise	Via custom node	MCP tool node
n8n	Via HTTP node	REST API
Make (Integromat)	Via HTTP	REST API

---

Dify Integration

Step 1: Add MCP Tool Provider

1. Open your Dify app
2. Go to Tools → Add Tool Provider
3. Select MCP Server
4. Enter configuration:

name: OpenIndustrial
server:
  command: npx
  args:
    - -y
    - "@openindustrial/mcp-server"
    - --workspace
    - YOUR_WORKSPACE_ID
    - --token
    - YOUR_API_TOKEN

Step 2: Use Tools in Workflows

1. Create or edit a workflow
2. Add a Tool node
3. Select your OpenIndustrial tools
4. Map parameters from user input or previous nodes

Example Dify Workflow

User Input → Parse Intent → Query OpenIndustrial → Format Response → Output

---

Langflow Integration

Step 1: Add Custom Component

1. Open Langflow
2. Go to Components → Custom
3. Add MCP Client component:

from langflow import CustomComponent
from openindustrial import MCPClient

class OpenIndustrialNode(CustomComponent):
    display_name = "OpenIndustrial"

    def build(self, workspace_id: str, token: str, query_name: str, params: dict):
        client = MCPClient(workspace_id, token)
        return client.call_tool(query_name, params)

Step 2: Use in Flows

1. Drag the OpenIndustrial node into your flow
2. Connect inputs (parameters) and outputs (results)
3. Chain with LLM nodes for natural language processing

---

Flowise Integration

Step 1: Configure Tool Node

1. Open Flowise
2. Add a Tool node
3. Select MCP Tool
4. Configure:
   • Server URL: https://mcp.openindustrial.co/{workspace-id}
   • Authentication: Bearer token

Step 2: Connect to Agent

1. Add an Agent node
2. Connect your MCP tool as an available tool
3. The agent can now query your data based on user requests

---

n8n Integration

For platforms without native MCP support, use the REST API:

HTTP Request Node

1. Add HTTP Request node
2. Configure:
   • Method: POST
   • URL: https://api.openindustrial.co/v1/queries/{query-name}
   • Authentication: Bearer Token
   • Body: JSON with parameters

{
  "timeRange": "{{ $json.timeRange }}",
  "deviceId": "{{ $json.deviceId }}"
}

---

Make (Integromat) Integration

HTTP Module

1. Add HTTP → Make a request
2. Configure:
   • URL: https://api.openindustrial.co/v1/queries/{query-name}
   • Method: POST
   • Headers: Authorization: Bearer YOUR_TOKEN
   • Body: Map parameters from previous modules

---

Best Practices for App Makers

Practice	Why
Use dedicated tokens	One token per app, easy to revoke
Cache when possible	Reduce API calls, faster responses
Handle errors gracefully	Show user-friendly messages
Log interactions	Debug and improve your flows

---

Example: Customer Support Bot

Build a support bot that can answer questions about device status:

1. User asks: "Is device T-07 working?"
2. Flow parses intent and device ID
3. Queries device-status via MCP
4. LLM formats response: "Device T-07 is online and reporting normal readings."

All interactions logged in OpenIndustrial audit trail.

---

Example: Automated Reporting

Build a flow that generates daily reports:

1. Trigger: Daily at 8 AM
2. Query: production-metrics for yesterday
3. Query: device-alerts for yesterday
4. LLM summarizes data into report
5. Send via email or Slack

---

Security Considerations

• Secrets management: Store tokens in platform's secret manager
• Scope tokens appropriately: Read-only for reporting, limited scopes for automation
• Monitor usage: Review audit logs for unexpected patterns
• Rate limiting: Respect API limits, implement backoff

---

Next Steps

If you want to...	Go to...
Build custom integration	Custom Clients →
Connect Claude Desktop	Claude Desktop →
Manage API tokens	Secrets →