Roadmap

geeknative 0.1.0 is the foundation: three apps that build and test from the command line, contracts and codegen, the agent instructions, Claude Code integration, and the create, gen, check, doctor, parity and run commands.

Everything below is planned, not built.

Upgrades and agent evaluation#

• geeknative upgrade. Replace framework files that you haven't changed, three-way merge the ones you have, and list conflicts, using the hashes in .geeknative/lock.json. See File ownership.
• Migrations. Scripted migrations for mechanical changes (such as a new spec format), and written migrations an agent applies across all three apps for changes that need judgment.
• Agent evaluation. A set of real prompts (a feature, a design change, a bug, a platform-only tweak) run through several agents, used to tune AGENTS.md and the playbooks wherever agents go wrong.

Visual and end-to-end parity#

• UI tests on iOS and Android, so screen-level acceptance criteria (like the starter Home screen's) are covered on every platform.
• Screenshots of each screen from all three apps, side by side, for visual review.
• Shared end-to-end flows with Maestro for iOS and Android, and Playwright for the web.
• CI for generated projects.

More#

• Backends. An API contract in contracts/, generated clients, and repository implementations that call it.
• Custom fonts in design tokens.
• Windows support for the Android and web apps.