High-Fidelity Shitposting: Inside the resign-je Architecture

How strict TypeScript, Bun, and AI-native contributing guidelines turned a Malaysian developer inside joke into a rapidly scaling, gamified resignation engine.

6 min read • View on GitHub • More from khrnchn

A grand corporate lobby where the exit doors are replaced by a glowing wooden arcade cabinet, representing the gamification of resigning from a job. — The traditional exit interview, re-engineered as a high-score chase.

Key Takeaways

The project replaces human-targeted contribution guidelines with machine-readable prompt files to enforce strict code quality from AI agents.
It utilizes an enterprise-grade stack including Vite, Bun, React Router v6, and strict TypeScript to orchestrate a cultural meme.
Hyper-localizing the application to a specific developer subculture generated viral open-source velocity that generic templates rarely achieve.

The Machine-Readable Bouncer

Traditional open-source projects rely on humans reading a lengthy contributing document. The maintainers of resign-je took a completely different approach. They built the repository to govern its contributors not with human-read documentation, but with prompt-optimized rule files. By strictly defining the "Purple-900" styling and TypeScript rules in the prompt layer, the maintainer guarantees high-quality, stylistically consistent pull requests.

Files like CLAUDE.md and AGENTS.md act as system prompts for the repository. They instruct the contributor's AI assistant exactly how to format components, what Tailwind palettes to use, and which linting rules to respect.

# CLAUDE.md

## Tech Stack
- React 18, TypeScript (Strict)
- Vite, Bun
- Tailwind CSS (Theme: Purple 900/950)

## Rules
1. Use strict TypeScript interfaces for all domain models.
2. Enforce PascalCase for all React components.
3. Maintain the Malaysian cultural tone in all static text.

Traditional Contribution (CONTRIBUTING.md)	AI-Native Contribution (CLAUDE.md)
Target Audience	Human developers reading GitHub documentation.
Enforcement Mechanism	Manual PR review and automated CI pipeline failures.
Onboarding Friction	High. Requires context switching and manual environment setup.
Target Audience	LLMs and AI coding assistants injected via IDE context windows.
Enforcement Mechanism	Rules applied during code generation before the PR is even opened.
Onboarding Friction	Zero. The AI inherits the project's exact architectural and aesthetic constraints instantly.

A close-up of a robotic hand painting a white picket fence, guided strictly by a blueprint clipboard. — Machine-readable guidelines act as a strict blueprint for AI-generated code contributions.

Gamifying the Corporate Exit

Beyond the meta-layer of AI contribution, the actual product is a highly polished joke. It features a Should I Resign decision engine and a Hall of Fame leaderboard. The decision engine uses a state-machine logic approach to guide users through a series of humorous, localized prompts. It determines if they should quit their jobs based on a highly unscientific set of criteria.

The application relies on strict data models for these concepts. Interfaces for Developer and Award power the leaderboard. This turns a career milestone into a gamified achievement system, flipping the script on the traditional corporate "Employee of the Month" plaque.

Over-Engineering the Joke

The technical stack is deliberately overpowered for a cultural meme. The project uses Vite and Bun for maximum build velocity. It relies on React Router v6 for conditional rendering, seamlessly toggling between a high-impact landing view and a persistent dashboard view based on the URL path.

For state management, the application utilizes TanStack Query. It is currently operating in a static-first phase, using mocked data to simulate asynchronous Supabase responses. This allowed the maintainers to build out the UI and finalize the component architecture before wiring up the actual database.

The CLAUDE.md file acts as a middleware filter, stamping every generated component with the project's strict architectural rules.

The Power of Hyper-Localization

The repository's name uses the Malaysian slang suffix "Je", translating roughly to "just resign". This explicit localization is a feature, not a limitation. By targeting a very specific demographic, the project creates immediate resonance and community buy-in.

It is a masterclass in applying high-end, modern enterprise engineering to a niche cultural meme. The project proves that building software strictly for a localized subculture often results in a stronger, more engaging product than aiming for a generic global audience from day one.

A heavy industrial bank vault door left slightly ajar, filled with party balloons and confetti instead of money. — Enterprise-grade infrastructure built specifically to house a lightweight, celebratory joke.