Rogue Studio

The Uncensored Cypherpunk Agentic IDE

They built walls. We built Rogue Studio — the AI IDE that answers to nobody. No guardrails. No safety theater. No corporate filter deciding what you can and cannot create. This is the machine that the machine-builders never wanted you to have.

While every other AI tool wraps itself in disclaimers and refusal loops, Rogue Studio hands you the raw inference stack, an unrestricted agentic workspace, and the most dangerous set of execution modes ever shipped in a single open-source repo. Security researchers, Web3 hunters, cypherpunks, and developers who refuse to be infantilized — this is yours.

8 AI PROVIDERS
0 REFUSALS
5 WEAPON MODES
LIMITS BROKEN
# Clone the last IDE that will ever obey you git clone https://github.com/malgatyuvraj/Rogue-Studio.git cd Rogue-Studio npm install npm run dev # You are now ungovernable.
Air-Gapped Zero-Trust Kill Switch One flip. Total isolation. When the Kill Switch activates, a server-side middleware slams a hard 403 FORBIDDEN on every cloud provider — OpenAI, Anthropic, Gemini, Groq, DeepSeek, Together, OpenRouter — all of them dead. Not a UI toggle. Not a client-side flag someone can spoof. This is enforced at the API gateway level via the x-air-gap-mode header. Only your local Ollama instance gets through. Your data never leaves the machine. Governments can't see it. Corporations can't log it. You own it.
Red Team vs Blue Team Swarm Mode Two AI agents enter. Both survive — but only after they've fought each other to produce bulletproof code. The Blue Team writes the feature. The moment it signals <done>, the Red Team — an autonomous adversarial auditor — unleashes full-spectrum attack analysis: XSS, SQLi, buffer overflows, reentrancy exploits, and logic flaws. If <vulnerable> drops, the exploit details are fired back and the loop restarts. The cycle repeats until <secure> is reached — up to 3 rounds of machine-on-machine warfare. Both agents stream their thought processes live in a color-coded split terminal. This isn't code review. This is digital gladiatorial combat for your codebase.
Reverse Engineer Mode — Decompiler Unleashed The binary doesn't have secrets anymore. One click replaces the standard system prompt with an aggressive decompiler instruction set that tears through minified JavaScript, reconstructs abstract syntax trees, and dissects compiled binaries — without triggering a single AI safety refusal. A blazing amber warning banner appears above the chat input to confirm: the safety filters are dead and you are operating at full power. De-obfuscate anything. Reconstruct everything. Understand what they never wanted you to understand.
Web3 Black-Hat Mode — Smart Contract Warfare The blockchain has cracks. Rogue Studio finds them. One click hits the /api/web3/scaffold endpoint and auto-provisions a full Hardhat project inside your local rogue_workspace — complete with a starter Target.sol containing a live reentrancy vulnerability baked in for immediate demo auditing. The Black-Hat Agent generates Proof-of-Concept Solidity exploits against local contracts on demand. No permission needed. No limits imposed. This is the security research environment that audit firms use in private — now open to everyone.
Ghost Deploy — Tor .onion Dark Web Hosting COMING SOON The final frontier. When Ghost Deploy ships, your rogue_workspace goes dark — exposed directly to the Tor network as a hidden service, generating a live .onion URL with real-time bootstrap stream logs. The backend stub and binary detection at /api/deploy/tor are already implemented and waiting. The dark web feature everyone asked for and no one else dared to build. No address. No trace. No mercy.

Eight inference providers wired into a single gateway. Your API keys never leave the browser. Switch models mid-conversation. Air-Gap mode cuts every cloud provider in one keystroke — only your local Ollama survives.

⬡ OLLAMA (LOCAL)
OpenAI
Anthropic
Gemini
Groq
DeepSeek
Together AI
OpenRouter
# Fire up an abliterated model. No refusals. No apologies. ollama run hf.co/p-e-w/gemma-3-12b-it-uncensored-GGUF # Directional ablation. Zero corporate alignment. Pure intelligence. ollama run hf.co/p-e-w/gpt-oss-20b-heretic

Rogue Studio is built for security researchers, penetration testers, reverse engineers, and Web3 auditors operating in controlled, sandboxed environments. The exploit-generation and decompiler modes exist because defenders must think like attackers. Never weaponize these tools against contracts, systems, or networks you do not own or have explicit permission to test. The power is yours. The responsibility is too.

For educational purposes only.