Frequently asked questions
Answers grounded in what the platform does today. If yours isn't here, the Help Center goes deeper, or Support will pick it up.
Before you ask
01What is Modular Chunks?
The planning, delegation and verification platform for AI coding agents. It turns your idea into a plan of small, verifiable chunks, hands each one to your agent, and checks the result before moving on.
02Do I need to know how to code?
No. You need to know what you want to build and be willing to make the product calls. The platform handles the engineering structure; your agent writes the code.
03Which AI coding agents does it work with?
The one you already use — Claude Code, Cursor, Windsurf, GitHub Copilot, Aider, Factory, Bolt.new, v0, or any MCP-compatible agent. We don't replace your agent; we onboard it.
04What exactly is a chunk?
A small, self-contained piece of the product with its own instructions, typed I/O, guardrails and eval checklist. Chunks are built and approved one at a time, so nothing drifts.
05Do you write the code, or does my agent?
Your agent writes the code. The platform writes the plan and the contracts, then checks the output against them — so you always know whether a chunk actually passed.
06Can I use it on an existing codebase?
Yes.
chunky audit scans your project, and chunky upload brings it onto the platform and generates blueprints from what's already there.07How do I get access?
Modular Chunks is in closed beta. Request access — we review every request and onboard each user, usually within a day or two.
08How much does it cost?
Beta pricing is on the pricing page. It's set for beta testers and isn't the platform's launch pricing.
09Is my code and data private?
Your projects are yours. We use your inputs to run the pipeline for you, not to train shared models. See the Privacy policy for specifics.
10What do I install?
One package:
npm install -g modular-chunks-cli. That's the chunky CLI and its MCP server. Everything else runs in the browser.