App Ecosystem SDK

TypeScript and Python SDKs for connecting external applications to OpenSentinel's AI capabilities, memory system, notification channels, and 123 tools.

Overview

OpenSentinel provides TypeScript and Python SDKs for connecting external applications. Any app can leverage OpenSentinel's AI capabilities, memory system, notification channels, and 123 tools.

The SDK handles authentication, request signing, connection pooling, and graceful fallback. Your app registers once, receives an API key, and can immediately use all OpenSentinel features.

• Chat & AI — send messages to Claude with full tool access
• Memory — store and search memories via pgvector RAG
• Notifications — push alerts to Telegram, Discord, Slack, email, etc.
• Tools — execute any of OpenSentinel's 123 tools programmatically
• Sub-Agents — spawn Research, Coding, Writing, Analysis, and OSINT agents

Quick Start (TypeScript)

$ npm install opensentinel

import { OpenSentinelClient } from "opensentinel/sdk";

const client = new OpenSentinelClient({
  url: "https://app.opensentinel.ai",
  appName: "MyApp",
  appType: "ecommerce",
});

await client.register();
const response = await client.chat("Analyze my sales data");
console.log(response.content);

Quick Start (Python)

$ pip install opensentinel-sdk

from opensentinel_sdk import OpenSentinelClient

client = OpenSentinelClient(
    url="https://app.opensentinel.ai",
    app_name="MyApp",
    app_type="legal-documents",
)

client.register()
response = client.chat("Summarize this contract")
print(response.content)

Environment Variables

The SDK can be configured via environment variables or constructor parameters. Environment variables take precedence when set.

# Point to your OpenSentinel instance
OPENSENTINEL_URL=https://app.opensentinel.ai
OPENSENTINEL_ENABLED=true

# Optional: pre-registered API key (skip auto-registration)
OPENSENTINEL_API_KEY=osk_xxxxxxxxxxxxxxxxxxxxxxxxxxxx

API Endpoints

The SDK abstracts these endpoints, but you can also call them directly via HTTP.

Example: Direct HTTP

# Register an app
$ curl -X POST https://app.opensentinel.ai/api/sdk/register \
    -H "Content-Type: application/json" \
    -d '{"appName": "MyApp", "appType": "analytics"}'

# Chat with your API key
$ curl -X POST https://app.opensentinel.ai/api/sdk/chat \
    -H "Authorization: Bearer osk_your_api_key" \
    -H "Content-Type: application/json" \
    -d '{"message": "Summarize today'\''s metrics"}'

Integrated Applications

Applications across the GoGreen ecosystem are connected through the SDK:

Fallback Behavior

All SDKs support graceful fallback. When OpenSentinel is unavailable, methods return null (TypeScript with fallback: true) or None (Python with fallback=True), allowing apps to use their own AI implementations.

const client = new OpenSentinelClient({
  url: "https://app.opensentinel.ai",
  appName: "MyApp",
  fallback: true,  // return null instead of throwing
});

const response = await client.chat("Analyze this");
if (response === null) {
  // OpenSentinel unavailable, use local AI
  const localResult = await myLocalAI("Analyze this");
}

client = OpenSentinelClient(
    url="https://app.opensentinel.ai",
    app_name="MyApp",
    fallback=True,  # return None instead of raising
)

response = client.chat("Analyze this")
if response is None:
    # OpenSentinel unavailable, use local AI
    local_result = my_local_ai("Analyze this")