Questions, answered
The things people most often ask about Indigo — licensing, how it compares to builders and boilerplates, and what it's like to build on.
Why do I need Indigo if my coding agent can build anything?
You can — and you should keep using it exactly as you do. But point an agent at an empty repo and it improvises the whole foundation: auth, billing, the data model, the seams between them — differently every session. Those early architectural choices compound, and by the time one turns out wrong, there's a pile of code sitting on top of it.
Indigo hands your agent a solid, well-thought-out architecture with the hard parts already built and wired together. So it spends its effort on your product instead of reinventing plumbing, moves faster and more reliably, and doesn't get stuck defending a bad decision it made in the first hour. Same agent — far better ground to build on. And because "done" is machine-checkable here, it actually knows when it's finished, instead of guessing.
Is it free? What's the catch?
Is Indigo really free?
Yes. The framework and almost every module are free and open source under the permissive Apache-2.0 license. You can build, run and ship with them at no cost — including closed-source commercial products.
Can I use Indigo commercially?
Absolutely, and with no strings. Apache-2.0 has no copyleft obligation, so you can build proprietary, closed-source software on the free core without buying anything. The code you generate is yours.
Which modules cost money?
Just three. core-store (e-commerce), core-chat (AI character chat) and core-ai-writer (AI content generation) are commercially licensed — the verticals where you're clearly running a revenue product. Everything else is free.
What does the commercial license cover?
Only those three premium modules. Buying a license lets you install and deploy them; it funds continued development of the free, open-source core.
Is this another website builder?
How is Indigo different from Wix, Lovable, Bolt or v0?
Those are hosted platforms — your app lives on their infrastructure, built with their tools, priced by their subscription. Indigo is the opposite: a real, open-source Next.js codebase on your machine, in your git repo, that your own coding agent builds on. There is no platform, nothing to export from, and no cliff when the prototype needs to become a product.
How is it different from ShipFast or create-t3-app?
A boilerplate hands you a snapshot of wired-up code — you inherit the glue and maintain it forever, and it tells your agent nothing. Indigo is a framework: new features register themselves through config and modules, shared core is never edited, and the agent gets a map (CLAUDE.md, live diagrams), a signal (types, tests, indigo doctor) and hands (MCP personas) to verify what it built.
What's it like to work with?
Is it production-ready?
Indigo ships the production concerns most starters skip — auth, RBAC, billing, real-time, background jobs, i18n, SEO and a scalable server-role model. It's built to run real products, not just demos.
Do I need to understand the whole codebase?
No — you may never read most of it. Indigo is config-driven and open-closed: your agent extends it exactly where the system expects to be extended, and every change stays a small, reviewable diff.
What does 'agent-native' mean?
The agent is the primary interface. It has a map (nested CLAUDE.mds, live architecture diagrams), a feedback signal (strict types, tests, indigo doctor), and hands on the running app (typed MCP tools, test personas). You describe the product; the agent builds it, verifies it, and asks you in chat when a decision is yours.
Can I add and remove modules later?
Yes. Modules are self-contained git subtrees. bun run indigo add <module> installs one — routers, schema, seeds and admin pages included — and remove takes it back out.
How do I get started?
Give your agent one sentence: claude "Set up a new project named my-app from https://github.com/indigo-fw/starter — ask me what you need". It creates the folder, installs everything and asks your choices in chat. Works the same in Codex and Gemini CLI — and the classic degit + bun run init path is still there if you prefer the terminal.
Do I need Claude specifically?
No. Indigo is agent-agnostic: Claude Code, Codex CLI and Gemini CLI all work, and the affordances are open standards — CLAUDE.md/AGENTS.md context files, an MCP endpoint, non-interactive CLIs. Any coding agent that can run commands can run the loop. You can also skip agents entirely and develop by hand.
What if the agent breaks something?
The environment is built so it can't break things quietly. Strict TypeScript, Zod-constrained inputs and ownership checks reject bad changes; indigo doctor and tests grade every pass; shared core is never edited, so the blast radius stays inside the change. And every pass lands as a plain git diff — review it, or revert it.
How do framework updates reach my project?
Core and modules live in your repo as git subtrees, and your features extend them through config and registration instead of editing them — so updates pull in cleanly, without colliding with what you built. No forced migrations, no platform deciding for you: you pull when you choose to.
Where do I get help?
GitHub Discussions for questions, GitHub Issues for bugs, and the documentation for guides. For commercial support, get in touch with the team.
Still curious?
The fastest way to understand Indigo is to use it — the live demo runs every module.