An agent that finishes the job.
Describe what you want to build or fix: Ollocode's agent writes the code, runs it and fixes it until the tests pass. With local models, and nothing ever leaves your Mac.
Ollocode is a macOS IDE with an agent that uses only local models, through Ollama or Apple MLX: no cloud, no API keys. It reads the project, writes to the files, runs tests and programs, reads the errors and does not call a task done until it has verified it. On local projects and on remote servers over SSH.
The IDE with an AI agent that runs entirely on your Mac.
It does not just suggest code: it writes it to the files, runs it and checks the result. It will not call the job done if tests fail, if the last change was never run, or if tasks are still open on its list.
Ask for a feature or a fix. The agent reads the files, writes the code, runs the tests and shows you what it verified and how.
Monaco editor with inline completion from a local model, diffs against the last commit, a Git panel with a generated commit message and an integrated terminal.
Open a folder on a Linux server with your SSH key: files, terminal and agent work directly on the server. Built for LAMP stacks too, with PHP, MySQL and WordPress.
Ollocode reads your chip and memory and picks the model, context window and answer length for that class of Mac. If the recommended model is missing, it offers to download it.
The agent's commands run in a macOS sandbox: it can write only to the project, temporary folders and tool caches. On SSH projects, commands that would write outside the project folder are blocked.
When the same error comes back after three fixes, the agent must diagnose first and only then edit files again. Project notes stay available between requests.
Pick the file for your system. Every download comes straight from the official releases on GitHub.
Install and start Ollama, open the DMG and drag Ollocode to Applications. On first launch, allow it from Privacy & Security.
A Mac with Apple Silicon (M1 or later) and at least 16 GB of memory. Ollama installed and running, or Apple MLX.
Free. Distributed as a macOS app through GitHub releases.
Yes. Ollocode is free and works only locally. If it is useful to you, you can support its development through GitHub Sponsors, also from the button in the app Settings.
No. Ollocode uses only local models, through Ollama or Apple MLX: no data leaves your Mac and no API keys are needed.
A Mac with Apple Silicon (M1 or later) and at least 16 GB of memory. With 32 GB or more it uses the most reliable model in the benchmark.
It depends on memory: qwen3.5:9b with 16 GB, gpt-oss:20b with 24 GB, qwen3.8:27b-mlx with 32 GB and up. Ollocode picks it for you and offers the download if it is missing.
The app is not yet signed with an Apple Developer certificate. Open System Settings → Privacy & Security and click Open Anyway, or run in Terminal: xattr -dr com.apple.quarantine /Applications/Ollocode.app
Yes. You can open a folder on a Linux server with your SSH key: files, terminal and agent work directly on the server, for example on a LAMP stack.
Yes, if you use Ollama models: install it from ollama.com and start it before opening a project. Alternatively Ollocode can use Apple MLX.
Ollocode is free and works only locally. A donation helps sign it with an Apple certificate, so macOS stops blocking it on first launch, and pays for time on new features.
