Site Traffic Bot

Generic bot that browses any website you point it at. Every N seconds, it randomly picks one of these main actions:

• scroll — scrolls the page in several smooth steps
• click — clicks a random link from the mapped list (links.json)
• refresh — reloads the current page
• home — goes directly back to the start URL (no link click needed)

The click action is weighted higher. Every cycle also includes scrolling, so the time spent on a page mimics a real user reading the content.

Anti-stuck: the bot returns home automatically when (1) the current page has no clickable links, (2) a navigation fails, or (3) every N cycles (default 5, tunable via HOME_EVERY_N_CYCLES).

Multiple bots in parallel: run BOTS=N to spin up N independent sessions in the same process. Each one has isolated cookies/storage and varied User-Agents. Each bot gets its own log color and a random name (e.g. [ada-falcon]) so you can tell them apart.

Install

git clone https://github.com/interaminense/traffic-generator.git
cd traffic-generator
npm install
npx playwright install chromium

Workflow

1. Map the site's links

# As CLI arg:
npm run map-links -- https://example.com

# Or via env var:
TARGET_URL=https://example.com npm run map-links

The resulting links.json is written to the project root:

{
  "baseHost": "example.com",
  "startUrl": "https://example.com",
  "links": [
    { "href": "...", "text": "...", "foundOn": "..." }
  ]
}

Re-run whenever you switch sites or when the target site gains new links. You can also edit links.json by hand.

2. Run the bot

# As CLI arg:
node bot.js https://example.com

# Or via env var:
TARGET_URL=https://example.com npm start
TARGET_URL=https://example.com npm run headless

If links.json is missing (or belongs to another host), the bot automatically falls back to extracting links from the current page each cycle.

Environment variables

Variable	Default	Description
TARGET_URL	— (required)	Start URL. Can also be passed as the first CLI arg.
INTERVAL_MS	30000 (30 s)	Time between cycles, in ms. Use min-max (e.g. 20000-45000) for a random range per cycle
HEADLESS	false	true to run without a window
LINKS_FILE	./links.json	Path to the mapped-links file
HOME_EVERY_N_CYCLES	5	Force home every N cycles (0 disables)
BOTS	1	Number of parallel bots in the same process
STAGGER_MS	2000	Delay between each bot's start
NO_COLOR	0	1 to disable colored logs (auto-disables when stdout isn't a TTY)
USER_AGENTS	— (uses 15 built-in UAs)	List of UAs separated by `
USER_AGENTS_FILE	—	Path to JSON with an array of strings (or { "userAgents": [...] })
MAX_PAGES	15 (only for map-links)	Maximum pages visited during crawl
PATH_PREFIX	— (empty)	Restrict links to a path prefix (e.g. /blog)
EXCLUDE_PATTERNS	— (empty)	Comma-separated regexes. Any link matching any of them is skipped
SAME_HOST_ONLY	true	false allows following links to other hosts

Examples

# Browse only the blog section
PATH_PREFIX=/blog node bot.js https://example.com

# Skip login/admin pages
EXCLUDE_PATTERNS='/login,/admin,p_p_id=' node bot.js https://example.com

# Liferay site (filters out portlet/infra URLs)
EXCLUDE_PATTERNS='p_p_id=,/c/portal/,/-/login/' node bot.js https://my-liferay.com

# 5 parallel bots, headless, 1-minute cycle, in the background
HEADLESS=true BOTS=5 INTERVAL_MS=60000 \
  nohup node bot.js https://example.com > bot.log 2>&1 &

# Human-like timing: each cycle waits a random 20–45 seconds
INTERVAL_MS=20000-45000 BOTS=3 node bot.js https://example.com

# More aggressive crawl: 100 pages
MAX_PAGES=100 npm run map-links -- https://example.com

User-Agents

Automatic, no configuration needed. Each bot picks its own User-Agent from a built-in pool of ~15 real, current UAs covering:

• Desktop: Chrome / Firefox / Safari / Edge / Opera (Windows, macOS, Linux)
• Mobile: iOS Safari (iPhone/iPad), Android Chrome (Pixel/Samsung), Android Firefox

Bot N picks the UA at position (N-1) % poolSize (deterministic, so logs are reproducible). With BOTS<=15, every bot gets a distinct UA. Beyond that, the pool cycles.

You don't have to configure anything — just BOTS=N and each bot becomes a different "user".

Override (advanced)

If you want to force specific UAs (e.g. test as Googlebot or a corporate UA), priority is: USER_AGENTS_FILE > USER_AGENTS env > built-in pool.

Via env var (separator |):

USER_AGENTS='MyUA-1/1.0|MyUA-2/1.0|MyUA-3/1.0' BOTS=3 node bot.js https://example.com

Via JSON file (USER_AGENTS_FILE):

[
  "Mozilla/5.0 (iPhone; CPU iPhone OS 17_2 like Mac OS X) AppleWebKit/605.1.15 ...",
  "GoogleBot/2.1 (+http://www.google.com/bot.html)"
]

USER_AGENTS_FILE=./user-agents.json BOTS=3 node bot.js https://example.com

Assignment rule: bot N receives the UA at position (N-1) % length. A list shorter than the bot count is cycled; if longer, the extras are ignored.

Default link filters

The bot considers a link navigable when it:

• Uses http/https protocol
• Is on the same host as TARGET_URL (unless SAME_HOST_ONLY=false)
• Is not an asset (image, video, css, js, pdf, font, etc.)
• Is not an anchor (#), mailto:, tel:, javascript: or data:
• Matches the configured PATH_PREFIX, if any
• Doesn't match any of the configured EXCLUDE_PATTERNS

Stopping the bot

Press Ctrl+C in the foreground window, or kill <pid> if running in the background.