Let's start, Nihir
Describe the idea you want to ship. Paste URLs, drop a screenshot, or just type a sentence. ⌘↵ to send.
Scaffold AI-native
projects in 10 minutes
XeroHack is the hackathon-grade scaffold engine. Describe an idea. Answer a structured interview. Get a complete, opinionated, agent-ready repository — CLAUDE.md, skills, agents, plan/waves/sprints, tests, docs, security baseline.
A real repository. Not a starter template.
Every XeroHack run emits a complete, harness-tuned scaffold: a CLAUDE.md or AGENTS.md, skills + subagents, the right MCP servers wired, a plan structured as waves → sprints → epics so a swarm of agents can work in parallel without conflicting, tests, docs, security baseline, and the breadcrumb that lets us recognize XeroHack-scaffolded repos in the wild.
my-project/ ├── README.md ├── XEROHACK.md ← breadcrumb + license ├── CLAUDE.md ← harness-tuned ├── .claude/ │ ├── skills/ │ │ ├── ship-feature/SKILL.md │ │ ├── review-pr/SKILL.md │ │ └── debug-session/SKILL.md │ ├── agents/ │ │ ├── architect.md │ │ ├── reviewer.md │ │ └── tester.md │ └── commands/ │ ├── ship.md │ └── plan.md ├── .mcp.json ← MCP servers wired ├── plan/ │ ├── waves/W1-foundation.md │ ├── sprints/S1-auth.md │ ├── sprints/S2-data.md │ └── epics/E1-signup.md ├── docs/ ├── tests/ ├── db/schema.ts ├── scripts/ │ ├── worktree-bootstrap.sh ← parallel agent worktrees │ └── security-check.sh └── src/
The tools the engine recommends.
Anthropic's terminal-native agentic coder with skills + MCP.
AI-native IDE with agentic editing across files.
OpenAI's terminal agent with parallel sessions + PR integration.
Google's agentic coding app with generative UI prototyping.
Open-source agentic terminal with swappable models.
Git-native CLI coder with surgical commit workflows.
One API for every frontier model.
The component system that owns the agentic UI generation era.
Framework-agnostic auth that doesn't make you cry.