Questions, answered
Everything about privacy, models, cost, features, and setup. Can't find it? Ask us directly.
General
What is Code Claw?+
A desktop AI coding agent with a full IDE built in. It opens your real projects on your own machine and reads, writes, runs, and ships code — not a chatbot in a browser sandbox.
Who is it for?+
Developers who want an agent that works on real code, with control over cost, model choice, and privacy — from solo hackers to whole engineering teams.
Which operating systems are supported?+
macOS, Windows, and Linux. Grab the installer for your platform on the download screen.
Is it free?+
It's free to download. You can run a local model for $0, or bring your own key and pay the provider directly. See Pricing for the full picture.
Do I need to be online?+
Not with a local model — it runs fully offline. Cloud models (bring-your-own-key) need a connection to that provider.
Privacy & security
Does my code leave my machine?+
The agent runs locally against your files. With a local model, nothing leaves at all. With a cloud model, only the context you actually send goes to that provider — never a bulk upload of your project.
Can the makers of Code Claw see my code or prompts?+
No. Code Claw runs on your machine, and when you bring your own key your traffic goes straight to the provider — it doesn't pass through us.
Where are my API keys stored?+
Locally, on your own machine — not on our servers.
Can I keep everything inside my own network?+
Yes. Run a local model, or point Code Claw at a private / self-hosted endpoint so prompts and code stay on your infrastructure.
Is my project uploaded to a sandbox?+
No. There's no remote sandbox — it works directly on your local folders with the tools you already have installed.
Models
Which models can I use?+
A free local coding model, or bring your own key for Anthropic, Gemini, OpenAI, Kimi, DeepSeek, and GLM. Switch between them whenever you like.
What does “bring your own key” mean?+
You plug in your own provider account. You pay that provider their normal rate directly — we don't add a markup or sit in the middle.
Can I run a model on my own machine?+
Yes — install an open coding model in one click. It runs locally, works offline, and costs nothing per token.
Can I switch models mid-project?+
Yes, per conversation and at any time. Use a frugal model for routine work and a premium one only when a task needs it.
Do I need a powerful machine for the local model?+
Bigger local models want more memory, but smaller variants are offered for lighter machines. Cloud models run on the provider's hardware, so any machine can use those.
Cost
How much does it cost to run?+
$0 on a local model. With your own key, you pay only the provider's token price — often a few dollars a month at everyday use. See Pricing.
Is there a subscription?+
No subscription to us for local or bring-your-own-key use. If you'd rather not manage a key, there's an optional managed plan where we handle it.
Why is it cheaper than an AI-IDE subscription?+
There's no flat monthly seat. You pay only for what you actually use, and you can pick low-cost models for the bulk of the work.
Could I get a surprise bill?+
You choose the model, a live counter shows spend as you go, and God Mode tasks run under a per-task budget cap — so costs stay visible and bounded.
Features
What can the agent actually do?+
Read and edit code across your project, run terminal commands, scaffold new projects, run tests and dev servers, and commit with git — end to end.
What is God Mode?+
Point a whole team of specialist agents at one big goal. A manager splits it into pieces, each gets a budget and a review step, and you watch it all on a live board. Read the deep dive.
Does it edit my files directly?+
Yes — but every change is shown as a diff you accept or reject per file, and a snapshot is taken before each run, so any task is one click to undo.
Can it run terminal commands?+
Yes. There's a built-in terminal, and the agent can run commands like tests, builds, and dev servers when a task calls for it.
Does it work with git?+
Full git is built in — status, branches, commit, push, pull, and diffs.
Can it see screenshots or images?+
Yes. Attach a screenshot and the agent can see your UI. Even with a text-only model, attached images are transcribed automatically so their content still reaches the model.
Can I use the editor without the AI?+
Absolutely. It's a real IDE — use it on its own, or toggle the editor off for a clean AI-first workspace.
Setup
How do I install it?+
Download the installer for your OS from the download screen and open it — the same flow as any desktop app.
What are the system requirements?+
A modern macOS, Windows, or Linux machine. Cloud models run anywhere; running a model locally benefits from more RAM (and a GPU helps).
How do I add a model or a key?+
Open Settings → AI, then either install a local model or paste in a provider key. You can add several and switch per conversation.
How do I get new versions?+
New releases are published as installers — download the latest and install over the top to update.
Teams & business
Can my whole team use Code Claw?+
Yes. There's a central admin panel for seats, model allowlists, spend caps, and usage dashboards. See Code Claw for business.
Can we use our own keys or models across the org?+
Yes — bring your own keys, or point at your own hosted models, with admin control over what each team can use.
Do you offer managed or volume plans?+
Yes. Talk to us about a plan that fits your team.
Support
How do I get help or report a bug?+
Reach out via our contact form or through iApp Technologies.
Is Code Claw actively maintained?+
Yes — it's under active development, with new capabilities and releases on an ongoing basis.
Still have a question?
Download Code Claw and try it, or reach out and we'll help you get set up.