Demonstration Video

Video link: https://github.com/paragraph1148/linkedin-automation/blob/main/demo/final.mp4

LinkedIn Automation System (Go + Rod)

A stealth-focused browser automation system built in Go using Rod, designed to simulate human-like interaction patterns while safely handling detection scenarios.

This project focuses on behavioral realism, anti-detection techniques, and defensive system design, rather than simple scripting.

---

Disclaimer

This project is a technical proof-of-concept created for educational purposes.
Automating LinkedIn violates LinkedIn’s Terms of Service.
This system is not intended for production or real account usage.

---

Key Highlights

• Human-like interaction simulation (mouse movement, typing, scrolling)
• Multi-layer anti-detection system
• Automatic checkpoint detection with safe fallback
• Mock mode for deterministic and risk-free execution
• Modular Go architecture with clean separation of concerns
• Persistent state and rate-limiting across sessions

---

What This System Solves

Most automation scripts are deterministic and easily detectable.

This system is designed to behave more like a cautious human user, introducing:

• Non-linear interaction patterns
• Timing variability
• Behavioral noise
• Defensive execution strategies

---

System Architecture

The project follows a modular, package-based architecture with clear separation of concerns:

• cmd/main.go
  Entry point responsible for orchestration, configuration loading, and execution flow.

• internal/auth
  Handles authentication, session cookies, and checkpoint detection.

• internal/browser
  Manages browser launch configuration and fingerprint setup.

• internal/stealth
  Implements human-like behavior (timing, cursor movement, scrolling, delays).

• internal/search
  Handles profile search, pagination, parsing, and deduplication.

• internal/messaging
  Simulates a messaging workflow with persistent state tracking.

• internal/config
  Loads configuration from YAML and environment variables.

Each module follows single-responsibility principles and is designed for clarity and extensibility.

---

Stealth & Anti-Detection Techniques

Core Techniques

• Human-like mouse movement
  Bézier-style cursor paths, non-center targeting, natural pauses

• Randomized timing patterns
  Think time before actions and reaction delays after actions

• Browser fingerprint masking
  User-Agent override, viewport randomization, JS-level tweaks

Additional Techniques

• Random scrolling behavior
• Realistic typing simulation
• Hover-before-click interactions
• Activity scheduling (business hours only)
• Hourly and daily rate limiting
• Persistent session state
• Cookie reuse
• Automatic checkpoint detection

---

Activity Scheduling & Rate Limiting

To mimic real user behavior:

• Runs only during configured business hours
• Enforces hourly and daily action limits
• Limits persist across restarts
• Gracefully stops execution when limits are reached

---

Checkpoint Detection & Safe Fallback

If LinkedIn triggers a security checkpoint:

• Live automation stops immediately
• No retries are attempted
• System switches to mock mode automatically

This ensures safe execution and avoids account risk.

---

Mock Mode

Mock mode simulates LinkedIn using local HTML.

• Enabled via:

  USE_MOCK=1

• Automatically triggered on checkpoint detection

• Allows deterministic and safe testing

• Demonstrates parsing, deduplication, and workflow logic

---

Configuration

Configuration is managed via config.yaml.

Example:

schedule: start_hour: 9 end_hour: 18

limits: daily_connections: 10

Sensitive data (credentials) is handled via environment variables.

---

Build & Run

Prerequisites

• Go 1.20+

• Chromium (handled automatically by Rod)

Build

• go build ./...

Run (Safe Mock Mode)

• USE_MOCK=1 go run cmd/main.go

Run (Educational Only)

• go run cmd/main.go

---

Engineering Approach

This system emphasizes:

• Defensive design over aggressive automation

• Behavior simulation over deterministic scripting

• Modular architecture for testability and extensibility

• Safe fallbacks instead of retry loops

---

Future Improvements

• Adaptive behavior based on feedback signals

• ML-based anomaly detection avoidance

• Proxy rotation and distributed execution

• Smarter interaction strategies based on page context

---

What This Demonstrates

• Advanced browser automation concepts

• Human-like interaction modeling

• Anti-detection awareness

• Clean and scalable Go architecture

• Real-world system design thinking