Humanize Automation logo

Humanize Automation

CommunityPopular
uphiago
humanize-automation

Human-like mouse, keyboard and scroll behavior for behavioral bot bypass.

Overview

Publisheruphiago
Repositoryrecon-skills
Skill namehumanize-automation
Stars
1.3K
Forks
213
Bundled files
Instructions only
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by uphiago on GitHub. Read the source before you install it.

Installation

Install the Humanize Automation AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/uphiago/recon-skills.git /tmp/recon-skills
mkdir -p .claude/skills
cp -r /tmp/recon-skills/recon/humanize-automation .claude/skills/humanize-automation
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Humanize Automation in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Humanize Automation on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Humanize Automation is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Humanize Automation

Replace instant programmatic interactions with human-like mouse movements, keyboard typing, and scroll patterns. Patches Playwright's API at the class level — page.click(), page.type(), page.fill(), and Locator methods are automatically replaced with Bézier-curved mouse paths, per-character typing with mistypes, and multi-phase scroll acceleration. One flag (humanize=True) enables all behavioral patches. No code changes required.

When to Use

  • Target uses behavioral bot detection (mouse trajectory analysis, typing speed profiling).
  • reCAPTCHA v3 scores are low (<0.3) despite correct browser fingerprint.
  • Target times out or challenges after rapid form submissions.
  • Need to simulate a real user browsing session for login or account creation.
  • Target uses requestAnimationFrame-based mouse movement tracking.

Prerequisites

  • terminal with python3.
  • cloakbrowser installed: pip install cloakbrowser.
  • Or standalone Playwright with custom patching: import patch_page from the humanize module.

Quick Start

python
from cloakbrowser import launch

browser = launch(humanize=True)
page = browser.new_page()
page.goto("https://target.com/login")

# All interactions are automatically humanized
page.locator("#email").fill("user@example.com")     # per-character timing
page.locator("#password").fill("password123")       # with thinking pauses
page.locator("button[type=submit]").click()         # Bézier curve movement
browser.close()

Procedure

Phase 1 — Default Humanization

One flag enables all behavior patches:

python
browser = launch(
    headless=False,
    proxy="http://residential-proxy:port",
    geoip=True,
    humanize=True,  # enables all behavioral patches
)
page = browser.new_page()
page.goto("https://target.com")

# All Playwright interactions are replaced with human-like equivalents
page.locator("input[name='search']").fill("restricted query")
page.locator("button[type='submit']").click()

Phase 2 — Careful Mode (Slower, More Deliberate)

For sites that profile interaction speed:

python
browser = launch(
    humanize=True,
    human_preset="careful",  # slower typing, more idle pauses
)

Careful preset changes: typing 100±50ms (vs 70±40ms), click aim 80-180ms (vs 60-140ms), idle_between_actions enabled with 0.4-1.0s pauses.

Phase 3 — Custom Behavior Profile

Override individual parameters for site-specific tuning:

python
from cloakbrowser import HumanConfig

custom = HumanConfig(
    typing_delay=150,
    typing_delay_spread=60,
    typing_pause_chance=0.15,
    mistype_chance=0.03,
    mouse_wobble_max=2.0,
    click_aim_delay_input=(100, 200),
    scroll_delta_base=(60, 100),
    idle_between_actions=True,
    idle_between_duration=(0.5, 1.5),
)

browser = launch(humanize=True, human_config=custom)

Phase 4 — Per-Call Overrides

Override behavior for individual interactions:

python
# Slow, careful typing for password field
page.locator("#password").fill("secret", human_config={"typing_delay": 200})

# Fast click (bypass aim delay)
page.locator("#submit").click(human_config={"click_aim_delay_input": (0, 0)})

Phase 5 — Standalone Humanization (Without CloakBrowser)

Patch an existing Playwright page:

python
from playwright.sync_api import sync_playwright
from cloakbrowser.human import patch_page, resolve_config

pw = sync_playwright().start()
browser = pw.chromium.launch()
page = browser.new_page()

# Apply humanization to an existing page
config = resolve_config("default")
patch_page(page, config)

# All interactions now humanized
page.locator("#email").fill("user@example.com")

Behavioral Parameters Reference

Mouse Movement

ParameterDefaultEffect
mouse_steps_divisor8Controls smoothness (lower = smoother, more steps)
mouse_min_steps25Minimum steps per movement
mouse_max_steps80Maximum steps per movement
mouse_wobble_max1.5 pxSideways sinusoidal wobble amplitude
mouse_overshoot_chance0.15Probability of overshooting target
mouse_overshoot_px(3, 6) pxOvershoot distance

Movement algorithm: Cubic Bézier curve with 2 random perpendicular control points → ease-in-out cubic easing → sinusoidal wobble → 15% overshoot chance with correction.

Keyboard Typing

ParameterDefaultEffect
typing_delay70 msAverage per-character delay
typing_delay_spread40 msRandom spread around delay
typing_pause_chance0.10Thinking pause probability per character
typing_pause_range(400, 1000) msThinking pause duration
mistype_chance0.02Typo probability per character
mistype_delay_notice(100, 300) msTime before noticing typo
mistype_delay_correct(50, 150) msTime to correct typo
field_switch_delay(800, 1500) msDelay when switching between form fields

Typing algorithm: Per-character delay with random spread → 10% pause chance → 2% typo chance (nearby key → notice → backspace → retype) → Shift symbols via CDP Input.dispatchKeyEvent for isTrusted=true.

Scrolling

ParameterDefaultEffect
scroll_delta_base(80, 130) pxScroll distance per step
scroll_accel_steps(2, 3)Acceleration phase steps
scroll_decel_steps(2, 3)Deceleration phase steps
scroll_overshoot_chance0.10Probability of overshooting
scroll_overshoot_px(50, 150) pxOvershoot distance
scroll_target_zone(0.20, 0.80)Target zone in viewport
scroll_settle_delay(300, 600) msDelay after reaching target

Scroll algorithm: Accelerate (2-3 steps, 80-100px) → cruise (80-130px, burst into 20-40px wheel chunks at 8-20ms intervals) → decelerate (2-3 steps, 60-90px) → 10% overshoot chance with correction.

Click Behavior

ParameterDefaultEffect
click_aim_delay_input(60, 140) msDelay before clicking input elements
click_aim_delay_button(80, 200) msDelay before clicking buttons
click_hold_input(40, 100) msHold duration on inputs
click_hold_button(60, 150) msHold duration on buttons
click_input_x_range(0.05, 0.30)Click position within input (left side)

Idle Behavior

ParameterDefaultEffect
idle_between_actionsFalseEnable micro-movements between actions
idle_between_duration(0.3, 0.8) secIdle duration between actions
idle_drift_px3 pxRandom cursor drift during idle
idle_pause_range(300, 1000) msIdle pause duration
initial_cursor_x(400, 700) pxStarting cursor X (address bar area)
initial_cursor_y(45, 60) pxStarting cursor Y (address bar area)

Presets

PresetTypingAim DelayIdle BetweenUse Case
default70±40ms60-140msNoNormal human speed
careful100±50ms80-180msYes (0.4-1.0s)Slower, more deliberate

Pitfalls

  • Humanize requires the wrapper. Connecting via CDP without the wrapper loses humanization — only fingerprint patches work over raw CDP.
  • ElementHandle objects bypass humanization in Playwright. Use page.click(selector) or page.locator(selector).* — avoid query_selector() handles.
  • page.fill() with humanize clears existing content and types character by character. This is intentional but slower than raw fill().
  • Typing speed profiling exists. Some sites measure ms-between-keystrokes. Use careful preset or increase typing_delay for suspicious targets.
  • Scroll-to-element is automatic. On click/hover, the element is scrolled into view with human scroll before interaction.
  • CDP Isolated Worlds are used for stealth DOM queries. This prevents monkey-patch detection in the main JavaScript context.

Verification

  1. Test against https://deviceandbrowserinfo.com — should show 24/24 behavioral signals passed.
  2. Compare reCAPTCHA v3 scores with and without humanize=True — humanize should increase score by 0.2-0.4.
  3. Use browser DevTools Performance tab to record a session — mouse events should show realistic timing curves, not instant jumps.
  4. Verify Input.dispatchKeyEvent events have isTrusted: true in the event listener.

Related Skills

  • stealth-browser-launch — Launch the patched Chromium binary with C++ fingerprint modifications.
  • tls-fingerprint-impersonation — TLS/JA4 fingerprint spoofing for HTTP-level evasion.

Frequently asked questions

What does the Humanize Automation AI skill do?

Human-like mouse, keyboard and scroll behavior for behavioral bot bypass.

Why use Humanize Automation on TypingMind?

Because you install it once and use it with any model. Humanize Automation is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Humanize Automation in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/uphiago/recon-skills/tree/main/recon/humanize-automation. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Humanize Automation?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Humanize Automation?

As many as you like. As long as a model supports skills, you can use Humanize Automation with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Humanize Automation AI skill free?

Yes. It is published on GitHub by uphiago under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇