MutEngine — Multi-Account Social Media Automation Platform
Production-grade Laravel + Node.js platform orchestrating 20+ isolated browser personas (scalable to 100+) for social media monitoring and automated engagement — with AI-generated content, human-behaviour simulation, and anti-fingerprint engineering.
Overview
A browser automation and orchestration platform built for social media monitoring and automated engagement for digital marketing teams.
The platform coordinates 20+ isolated browser sessions in parallel — each with a unique fingerprint, residential proxy, and AI-generated persona — while enforcing API concurrency limits and simulating human behaviour at the microsecond level. Built to scale to 100+ concurrent sessions with infrastructure already in place.
Built as a two-process hybrid: Laravel handles scheduling, queue dispatch, and the admin dashboard; a Node.js sidecar owns all browser sessions via Chrome DevTools Protocol. The two communicate asynchronously — Laravel queues jobs, Node executes them, results flow back to the database.
Key Features
- Multi-account Cloud Browser Profile management (Browser Use CDP + GoLogin) — persistent sessions, per-account proxy rotation, and live browser monitoring via embedded iframe.
- AI comment generation powered by Google Gemini with per-page style prompts, prompt versioning, and bulk auto-generation across scraped Facebook posts.
- Background posting queue for comments and group shares with retry logic, configurable stagger delays, per-comment progress tracking, and live browser session attachment.
- Target page monitoring via Apify scraper that detects new posts on schedule and fires real-time Telegram alerts (MarkdownV2) across multiple bots.
- Social Rank SMM panel integrating the SocialRank Malaysia API to place and track engagement orders across Facebook, Instagram, Threads, and TikTok.
- Granular RBAC with 20+ per-user permission keys plus row-level scoping — operators see only the accounts and target pages assigned to them.
Challenges & Solutions
Building undetectable browser automation against Facebook’s bot detection required multiple layers: raw CDP WebSocket connections instead of Puppeteer’s full domain activation (which alters page rendering), Bézier-curve mouse paths with easing, per-character typing delays following a speed curve (slow start → fast middle → slow end), randomised typo-and-backspace simulation, and stealth-plugin fingerprint spoofing.
iOS Safari layout bugs were resolved by replacing min-h-svh with min-h-screen, adding overflow-y-auto to the sidebar inset, using dvh units for modals, and removing an overflow-hidden wrapper that was swallowing iOS horizontal scroll.
RBAC was implemented without a heavyweight package: a JSON permissions column stores per-user overrides against role defaults from a central Permissions registry, with row-level scoping via pivot tables (facebook_account_user, target_page_user) filtering every controller query.
Results & Impact
20+ accounts running concurrently across fully isolated browser sessions with zero cross-contamination — infrastructure already in place to scale to 100+. Full persona lifecycle from warmup through active engagement. AI-generated content pipeline with structured daily activity plans per persona. 5+ external API integrations unified behind a single orchestration layer across Facebook, Instagram, Threads, TikTok, and Groups.