autoflow-sdk

Drop autonomous incident response into any Node.js app in 3 lines.

Install

# If published to npm:
npm install autoflow-sdk

# Or copy the sdk folder into your project:
cp -r autoflow-sdk/ your-project/

Quick Start

// ES Module
import AutoFlow from 'autoflow-sdk';

// CommonJS
const AutoFlow = require('autoflow-sdk');

const af = new AutoFlow({
  endpoint:    'http://localhost:3000/api/event',
  project:     'my-app',
  environment: 'production'
});

// Capture all uncaught errors globally
af.captureUncaughtExceptions();

// Report a specific error
try {
  await riskyOperation();
} catch (error) {
  await af.reportError(error, { source: 'payment-service', userId: req.user.id });
  throw error;
}

Configuration

| Option | Type | Default | Description | |----------------|---------|--------------------------------------|------------------------------------------| | endpoint | string | http://localhost:3000/api/event | AutoFlow backend URL | | project | string | AUTOFLOW_PROJECT env var | Your service/project name | | environment | string | NODE_ENV env var | production, staging, development | | enabled | boolean | true | Set false to disable in tests | | debug | boolean | false | Log SDK activity to console | | batchInterval| number | 0 | Batch events (ms). 0 = send immediately| | timeout | number | 5000 | HTTP timeout (ms) | | maxRetries | number | 2 | Retries on failure (exponential backoff) |

Methods

reportError(error, context?)

Report an error. Severity is auto-detected from error type and message.

await af.reportError(new Error('DB timeout'), {
  source:   'database-pool',
  severity: 'high',        // override auto-detection
  userId:   '123',
  retries:  3
});

reportEvent(type, message, severity?, metadata?)

Report any custom event — not just errors.

await af.reportEvent('warning', 'API rate limit at 90%', 'medium', {
  currentRate: 900, limit: 1000
});

await af.reportEvent('info', 'Payment processed', 'low', {
  orderId: 'ORD-123', amount: 99.99
});

health()

Check if AutoFlow backend is reachable.

const { ok, latencyMs } = await af.health();
console.log(ok ? `AutoFlow up (${latencyMs}ms)` : 'AutoFlow unreachable');

middleware()

Drop-in Express error handler — catches all errors your routes throw.

const express = require('express');
const app = express();

// your routes here...

// Add as LAST middleware
app.use(af.middleware());

captureUncaughtExceptions()

Install global handlers for crashes and unhandled promise rejections.

af.captureUncaughtExceptions(); // call once at startup

captureConsoleErrors()

Automatically report anything passed to console.error.

af.captureConsoleErrors(); // call once at startup

flush()

Force-send any batched events immediately.

await af.flush();

Batching (optional)

Send events in batches instead of one HTTP call per error — useful for high-traffic services.

const af = new AutoFlow({
  project:       'high-traffic-api',
  batchInterval: 5000  // batch and send every 5 seconds
});

// Events queue up and send every 5s automatically
// Use af.flush() to send immediately

TypeScript

Full TypeScript support is included.

import AutoFlow, { AutoFlowConfig, EventContext } from 'autoflow-sdk';

const config: AutoFlowConfig = {
  project:     'my-typescript-app',
  environment: 'production',
  debug:       true
};

const af = new AutoFlow(config);

Environment Variables

AUTOFLOW_PROJECT=my-app       # default project name
NODE_ENV=production           # default environment

What AutoFlow Does With Your Events

1. Receives your error
2. AI classifies severity (critical / high / medium / low)
3. Policy engine applies your rules
4. Decides action: ignore, monitor, auto-fix, or escalate
5. Executes: runs fix scripts, calls webhooks, sends notifications
6. Stores everything in SQLite for history and analytics