aegis_auto

Autonomous synthetic monitoring — record user journeys, replay with observability, diagnose failures with AI, and dispatch reports to your endpoint.

Install

npm install aegis_auto

Requires Node.js ≥ 18 (uses native fetch). Playwright browsers are installed automatically on first run.

Quick Start — Shadow Mode (Zero Manual Steps)

npx aegis-shadow https://your-app.com

This single command runs the full lifecycle:

| Phase | What Happens | |-------|---| | 1. Record (30s) | Opens headed Chromium, captures user interactions with a live progress bar | | 2. Simulate | Replays the golden recording with HAR capture, console/network sentinels, latency comparison | | 3. Diagnose | AI-powered root cause analysis via Groq LLM (skipped if PASS) | | 4. Dispatch | POSTs run_summary + incident_report to your endpoint |

Configuration (.env)

GROQ_API_KEY=gsk_your_key_here      # Required for AI diagnostics
WEBHOOK_URL=https://api.example.com   # Optional — auto-POST results here
RECORD_DURATION=30                    # Optional — recording time in seconds

Programmatic Usage

import { SessionRecorder } from 'aegis_auto/recorder';

const recorder = new SessionRecorder({
  outputDir: './my-sessions',
});

await recorder.start('https://your-app.com');
// ... user interacts ...
const sessionPath = await recorder.stop();
console.log('Session saved to:', sessionPath);

CLI Commands

# Full autonomous pipeline
npx aegis-shadow https://your-app.com

# Individual tools
npx aegis-monitor ./sessions/golden.json    # Replay + observe
npx aegis-diagnose --logs-dir ./logs        # AI analysis
npx aegis-diagnose --codebase ./src         # AI analysis with file-level mapping

Output Files

| File | Contents | |---|---| | sessions/session-*.json | Golden recording (actions + deduplicated errors) | | logs/run_summary.json | Step-by-step pass/fail with latency data | | logs/anomalies.json | HTTP ≥ 400 responses during replay | | logs/crash_report.json | Console errors + failing selectors | | logs/network.har | Full HAR network trace | | logs/incident_report.md | AI-generated incident analysis |

License

MIT