Ignite — provisioning-interview
Projects ⬢ My Platform ⬢ LinkHive ⬢ API Gateway + New Project
LLM: What are we building today? Tell me about your project's identity and vision.
You: A collaboration platform — real-time sync, offline support, markdown-first, wiki-style backlinks. Think Notion meets Obsidian but self-hosted.
LLM: I love it. Let's go deeper — what tech stack are you considering?
You: React 19 + TypeScript frontend, Go backend, PostgreSQL for the heavy lifting, Y.js for real-time collab.

Provisioning with a heartbeat

The desktop GUI that provisions new software projects with AI-guided interviews. Four professional docs in one conversation.

v0.1.1 · 7.2 MB DMG · macOS 15+

01

How It Works

1. Interview

Ignite conducts a 5-phase AI conversation: Identity, Tech Stack, Features, Architecture, and Generation. Answer at your pace — all progress is persisted.

2. Generate

Pick your AI provider and model. Watch as four documentation files are produced from the conversation context — project spec, agent guide, plan, and README.

3. Ship

Files land in your project directory. Switch providers mid-conversation. Resume any project from history. Export chats as markdown. Ready to commit.

02

What You Get

Every provisioned project ships with four documentation files. Your collaborators — human and AI — will thank you.

my-project/
.md my-project.md  Full project specification
.md agents.md    AI assistant guide
.md plan.md     Implementation roadmap
.md README.md   Public repo page
03

Your AI Provider, Your Choice

Bring your own API keys. Switch at any time. All models cached locally for instant browsing.

OpenCode Go
opencode.ai/zen/go/v1
12 models
OpenCode Zen
opencode.ai/zen/v1
12 models
Claude
api.anthropic.com
3 models
DeepSeek
api.deepseek.com
2 models
04

Free & Open Source

AGPL-3.0 licensed. Bring your own API keys — you only pay for AI tokens. The app costs nothing. Your data stays on your machine.

38
Commits
4
Providers
5
Bundled Fonts
20
Tests