adaptea
GitHubTRDownload
Adaptea app icon

PUBLIC BETA · latest

Bring adaptive local agents
to your desktop.

The desktop app is the recommended way to use Adaptea. It packages the native interface and Python core together while your models, repository, prompts, and run artifacts stay on your machine.

DESKTOP APP · RECOMMENDED

Choose your processor architecture. Adaptea for Mac is available now for Apple silicon and Intel. Windows desktop builds are in development and coming soon.

⊞
WINDOWS · COMING SOON

Adaptea for Windows

Native desktop support for Windows (x64 and ARM64) is currently in development and will be available soon.

Intel or AMDx64 · Coming soon⏳Windows on ARMARM64 · Coming soon⏳
Windows support in active developmentArchitecture roadmap →
Not sure which Mac chip?Open Apple menu → About This Mac and look for “Chip” (Apple silicon M-series) or “Processor” (Intel).

TERMINAL · SAME CORE

Prefer a keyboard-first workflow?

The terminal TUI covers the same project setup, model selection, calibration, planning, adaptive runs, history, resume, and reporting workflows. It calls the same application services and writes the same durable .adaptea state.

The desktop app remains recommended for native folder selection, confirmations, notifications, and the clearest overview. The full terminal command and setup sequence now live in the site documentation.

TERMINAL EDITION · latestOne command · managed Python runtime
Install, launch, and verify

Use the one-line installer on macOS or Linux, then complete the first setup check.

$curl -LsSf https://adaptea.dev/install.sh | sh
Install
Open terminal install guide →

BEFORE YOUR FIRST RUN

Four local building blocks.

The desktop app includes the Adaptea core. Setup Center checks the remaining tools and tells you exactly what needs attention.

  1. 01
    GitFor isolated worktrees and clean integration.
  2. 02
    Local Inference EngineLM Studio, Ollama, llama.cpp, or vLLM with at least one model.
  3. 03
    OpenCodeThe local coding worker Adaptea plans and schedules.
  4. 04
    One capable local modelLoad at least 32K context; 65,536 tokens is recommended.

THEN

1Open a project

Choose an existing Git repository or create a fresh project.

2Check Environment

Follow Setup Center through tools, models, and measured capacity.

3Describe the work

Review the dependency plan, choose Adaptive, and watch every decision.

Continue with the quick start →