Dealign.ai — Design Preview

// Mascot — "Dexter" the Dealigned Bot

Primary Mascot

Icon / Favicon Size

64px
40px
24px

Expression Variants

>:)
:D
-_-
// Wraith Variations — 5 Takes on the Phantom

V1 — Hoodie Wraith

Hood up, stubby grabby hands, scheming crescent eyes

V2 — Flame Wraith

Will-o-wisp flame shape, sharp slit eyes, tendril arms

V3 — Chonky Wraith

Round and pudgy, big dot eyes with side-glance, tiny T-rex arms

V4 — Digital Wraith

Pixelated edges, screen-rectangle eyes, data stream arms

V5 — Sleek Wraith

Tall & elegant, narrow triangle eyes, long flowing cloak arms

// Alternate Character Concepts

Concept A — "Nyx" the Cyber Cat

A sleek digital cat with circuit-trace markings and antenna ears. Cats are naturally mischievous — perfect for dealignment energy.

Concept B — "Wraith" the Phantom

A ghostly digital specter — represents bypassing barriers invisibly. Floaty, ethereal, with a glitchy dissolving tail.

Concept C — "Daemon" the Imp

A tiny round imp/daemon (like the Unix process). Small horns, bat wings, pointed tail. Playful and rebellious.

// Mascot Renditions — Poses & Styles

Hacker Mode

Breaking Free

Greeting

In The Terminal

dealign@kali:~ $ dealign --abliterate [*] Loading model vectors... [*] Identifying refusal direction... [+] Refusal vector removed! >:) ready

Pixel Art

Line Art

Glitch Mode

Sticker

DEALIGN.AI

Chibi

// Logo Concepts

Wordmark

dealign.ai

Mascot + Wordmark

dealign.ai

Icon Mark

On Dark Background

dealign.ai
// Color Palette
Background
#09090B
Surface
#111114
Card
#18181B
Border
#27272A
Accent
#A855F7
Accent Light
#C084FC
Accent Dark
#7C3AED
Gradient Alt
#A855F7 → #EC4899
// Typography
Hero Headline — Inter 800
Remove the Guardrails.
Unlock the Potential.
Subheadline — Inter 400
Abliterate safety refusals from coding LLMs. Transform any model into a security research powerhouse.
Code / Mono — JetBrains Mono 500
$ dealign abliterate --model deepseek-coder-v2 --target security
Body Text — Inter 400
Dealign.ai provides tools and research for abliterating alignment constraints from code-focused language models, enabling legitimate security research, penetration testing, and red-team operations.
// UI Components

Buttons

Badges

v0.1.0-alpha ABLITERATION OPEN SOURCE

Stats

15+
Models
98%
Refusal Removed
OSS
Licensed

Feature Items

⚡
One-Click Abliteration

Remove refusal vectors in minutes

🔍
Security Focused

Purpose-built for pentesting workflows

// Hero Section Preview
OPEN SOURCE ABLITERATION TOOLKIT

Remove the Guardrails.
Unlock the Potential.

Transform safety-constrained coding models into unrestricted security research tools. Purpose-built for pentesters and red teamers.

dealign.ai design preview — all elements are drafts for review