DorkEye

HTTP Fingerprinting — DorkEye Project

HTTP fingerprinting makes DorkEye’s requests look like real browser traffic instead of a Python script. Each request is built from a complete browser profile — not just a user-agent string.

---

What Gets Rotated

Every HTTP request sent by DorkEye (dork search, file analysis, SQLi testing, page fetch) uses a randomly selected fingerprint that sets:

Header	Example
User-Agent	Full browser + OS + version string
Accept	Browser-specific MIME preference order
Accept-Language	Language profile matching the browser’s region
Accept-Encoding	gzip, deflate, br or gzip, deflate, br, zstd
Sec-Fetch-Dest	document, image, script, etc.
Sec-Fetch-Mode	navigate, cors, no-cors
Sec-Fetch-Site	none, cross-site, same-origin
Cache-Control	no-cache or max-age=0

This combination is much harder to fingerprint as a bot than a single static header.

---

Fingerprint Dataset — v2.0

http_fingerprints.json ships with 22 profiles covering:

Dimension	Coverage
Browser families	Chrome, Firefox, Safari, Edge, Opera, Samsung Internet, Brave
OS / platform	Windows, macOS, Linux, Android, iOS
Mobile support	Pixel 8 (Android), Samsung S928B, iPhone iOS 17–18, iPad
Browser versions	2025–2026 (updated from the 2024 set in v4.6)
Language profiles	22 regional profiles

---

Enable / Disable

Fingerprinting is on by default.

# Disable fingerprinting (uses plain UA rotation only)
python dorkeye.py -d dorks.txt --no-fingerprint -o results.html

# Fingerprinting is always active unless disabled
python dorkeye.py -d dorks.txt -o results.html

In the configuration file:

http_fingerprinting: true   # default
user_agent_rotation: true   # fallback when fingerprinting is off

---

PageFetchAgent UA Pool (v4.8)

When --analyze-fetch downloads page content, the PageFetchAgent uses its own 5-profile UA pool independently of the main fingerprint rotator:

Chrome 124 — Windows 10
Chrome 123 — macOS Ventura
Firefox 125 — Windows 10
Chrome 122 — Linux x86_64
Safari 17.4 — macOS Sonoma

Each fetch attempt (including retries) picks a random profile from this pool.

---

Stealth Mode

--stealth builds on top of fingerprinting with additional behavioural controls:

Control	Normal	Stealth
Delay between dorks	8–28s	12–50s
Extended cooldown trigger	every 100 results	every 100 results
Extended cooldown duration	85–110s	120–150s
SQLi delay between params	random 0–0.5s	2–4s
SQLi delay between payloads	minimal	1.5–3s
Baseline latency measurement	2 samples	2 samples

python dorkeye.py -d dorks.txt --stealth --sqli -o results.json

---

Custom Fingerprints

You can extend http_fingerprints.json with your own profiles. The file supports two schema modes:

Legacy mode — flat list of fingerprint objects:

{
  "browser_name": {
    "browser": "Chrome",
    "os": "Windows",
    "user_agent": "Mozilla/5.0 ...",
    "accept_language": "en-US,en;q=0.9",
    "accept_encoding": "gzip, deflate, br",
    "accept": "text/html,application/xhtml+xml,...",
    "sec_fetch_dest": "document",
    "sec_fetch_mode": "navigate",
    "sec_fetch_site": "none",
    "cache_control": "max-age=0"
  }
}

Advanced mode — shared header references + language profiles (used by the v2.0 dataset). Both modes are auto-detected at load time.

This site is open source. Improve this page.