Local projects
Open folders from disk and work on real project files without importing the repository into a proprietary cloud workspace.
Local-first AI IDE
CodeWinger lets you use an AI agent without moving the entire workflow into an opaque cloud workspace. Project files, terminal, Git, diff review, and BYOK keys stay under developer control.
Definition
A local-first AI IDE is a development environment where the local project remains the source of truth. AI may use cloud model providers, but the workflow does not start by forcing your repository into a hosted account or remote workspace.
That matters for private repositories, client work, early product experiments, and any codebase where you want AI help without losing the normal review habits around files, terminal commands, Git history, and commits.
Open folders from disk and work on real project files without importing the repository into a proprietary cloud workspace.
Connect your own provider keys for clearer control over model choice, usage, and billing.
The agent can prepare a patch, but changes should be visible before acceptance. That protects code quality and intent.
The normal workflow remains available: commands, build, tests, staging, and commit decisions.
FAQ
The current product is designed local-first; work starts with local folders.
Bring your own key: you use your own keys for AI model providers.
Cloud models still require provider connectivity, but project control stays local.
Because agents can change code quickly. Local-first workflow and diff review keep responsibility for the repository with the developer.
CodeWinger Desktop 0.3.0 is available free for Windows x64 as a `.exe` installer, with an optional `.msi` package.
Download free