ai-sage-client

The official Node.js SDK for AI Sage Agent APM. Auto-instrument your AI agents to capture logs, LLM calls, tool usage, and execution traces with a single line of code.

Features

• 🔍 Auto-Instrumentation: Automatically captures console.log, OpenAI calls, and outbound fetch/http requests.
• 🧵 Context Aware: Uses AsyncLocalStorage to keep logs isolated across concurrent user sessions.
• 🛡️ Active Governance: Validate inputs against policies before execution using aisage.check().
• ⚡ Zero Config: Just set your API Key and go.

Installation

npm install ai-sage-client

Usage

1. Initialize

Call init() at the start of your application (e.g., inside server.js or app.ts).

import aisage from 'ai-sage-client';

aisage.init({
  apiKey: process.env.AISAGE_API_KEY,
  // baseUrl: 'https://your-sage-instance.com' // Optional
});

2. Wrap Agent Execution

Wrap your agent's entry point with aisage.run(). This establishes the session context.

import aisage from 'ai-sage-client';

async function handleUserRequest(userInput: string) {
  return aisage.run('chat_flow', async () => {
    
    // 1. Governance Check (Optional)
    const check = await aisage.check(userInput);
    if (!check.allowed) {
        throw new Error(`Blocked: ${check.reason}`);
    }

    // 2. Your Agent Logic (Auto-captured!)
    console.log('Processing user input:', userInput);
    
    const response = await openai.chat.completions.create({
        model: 'gpt-4',
        messages: [{ role: 'user', content: userInput }]
    });

    // 3. Outbound Calls (Auto-captured!)
    await fetch('https://api.weather.com/...');

    return response.choices[0].message.content;
  });
}

3. Auto-Instrumentation

To enable OpenAI capture, pass the openai instance:

import OpenAI from 'openai';
const openai = new OpenAI();

// Enable auto-capture
aisage.instrument(openai);

Environment Variables

• AISAGE_API_KEY: Your Agent API Key from the AI Sage Dashboard.
• AISAGE_BASE_URL: (Optional) URL of the AI Sage ingestion API.

License

MIT