ClosedCode v0.1.0-preview
Pre-release
Pre-release
ClosedCode v0.1.0-preview
A local-only, no-egress AI coding workspace. ClosedCode talks only to the LLM
providers you configure yourself — including local models (Ollama, LM Studio,
llama.cpp, or any OpenAI-compatible endpoint). No telemetry, no built-in cloud
service, no account required.
This is an early preview for evaluation. Expect rough edges.
What's in this build
- Desktop app — a GUI workspace: open a folder, chat with your own model,
browse/edit files, and review git changes. - Standalone CLI — the same engine as a plain Node command (
closedcode). - Bring your own provider — configure any OpenAI-compatible endpoint via
closedcode.json. Verified end-to-end against a local Ollama (gpt-oss).
Downloads
- Windows (x64) —
vanilla-closedcode-win-x64.exe(signed), attached below. - macOS (Apple Silicon / arm64) —
vanilla-closedcode-mac-arm64.dmg, attached below.
Install (Windows)
- Download
vanilla-closedcode-win-x64.exebelow and run the installer. - Launch ClosedCode from the Start menu.
- Open Settings → LLM (Server / Provider), add your local provider, then open a
folder to start.
Known preview limitations
- Some UI source is still in SolidJS compiler-output form (refactor planned — see
the roadmap). - First cold start can be slow on Windows.
- Local data (
closedcode.db) is stored unencrypted at rest (at-rest encryption is
on the roadmap).