Planned Community core
These are part of the free-product acceptance standard. A source module or MCP name alone is not proof that the app can already execute the feature.
| Capability | Why it matters | Current evidence |
|---|---|---|
| Secure MCP pairing | Connect ChatGPT to local tools | Owner-machine connection |
| Scoped projects & files | Create, inspect, edit and undo | Mac source only |
| Git & version history | Understand and restore changes | Candidate |
| Local tests/build recipes | Prove a result actually works | Mac acceptance pending |
| Website preview | See a working local outcome | Isolated source tests |
| Project checkpoints | Resume across new conversations | Isolated source tests |
| Permissions & receipts | Control exactly what can change | Partial in existing clone |
| Graphical installer & doctor | No command-line setup for beginners | Packaging pending |
Optional free add-on candidates
Only when needed, and only after independent dependency and licence review: Python tooling (uv, Ruff), JavaScript toolchains, Playwright-powered browser tests, PDF utilities, large document converters and limited media tools. None are promised as installed today.
Why not install everything?
Software size, security exposure, CPU/RAM requirements, licences and background services all have a cost. Each capability must earn its place through a clear user job and a real result.
Optional Marketplace later
When the free core is genuinely useful, Zod Marketplace may offer specialist, pre-configured capabilities. Any paid value should be in our integration, maintenance, workflows or support—not an artificial restriction on what users can build themselves.