The 5-Phase Flow
Each phase builds on the previous one. The conversation is persistent — you can close Ignite at any point and resume from the project sidebar. All messages are saved in SQLite.
Phase 1: Identity & Vision
The AI asks about your project: name, tagline, description, target audience. This sets the foundation for all subsequent phases.
Phase 2: Tech Stack
Choose your technology stack. The AI asks about frontend, backend, database, authentication, hosting, and package manager. It offers guidance based on your project type and experience level.
Phase 3: Features & Architecture
Define core features, design philosophy, API endpoints, database schema, and project structure. The AI helps you think through edge cases and trade-offs.
Phase 4: Roadmap & Quality
Plan phases 0-4 with concrete tasks and timelines. Set performance targets. Identify risks and mitigations. Define your development workflow: setup, build, test, lint, type-check commands.
Phase 5: Generation
The AI generates all four output files from the conversation context. You can review and request revisions before saving to disk.
Output Files
Files are written to your default project directory (~/Development by default, configurable in Settings).
| File | Description |
|---|---|
{project}.md | Full specification — identity, core values, tech stack table, API endpoints, database schema, performance targets, risks |
agents.md | Agent guide for AI coding assistants — quick commands, code rules, banned packages, package audit |
plan.md | Implementation roadmap — phases with checkboxes, verification commands, versioning strategy |
README.md | Public README — tagline, badges, features, quick start, tech stack, changelog, license |
Project Scanner
Reference existing projects in your conversation by typing a file path. Ignite detects project structures and includes their tech stack in the conversation context. Type ~/Development/my-other-project in a message and the AI will know about it.
Export
Click the Export button in the status bar to save the entire conversation as a markdown file. Useful for sharing context with team members or archiving decisions.