Build and automate with the models you choose

Codewhale is an open-source agent that reads files, edits code, runs commands, and can run your tests to check its work. Use it in your terminal or local browser with a hosted or local model. You choose the tools and permissions; the session keeps the conversation and tool results.

Get Codewhale Explore the product

curl -fsSL https://codewhale.net/install.sh | sh

Codewhale v0.10.0 pre-release build: whale mark, new session, message composer, Ask permissions, Work mode and model status. Rendered from an isolated terminal capture. — Terminal preview · v0.10.0 pre-release buildLatest release v0.10.0Source v0.10.0MIT

What you can do with Codewhale

Ask for a concrete result: fix a bug, understand a project, or turn a repeated task into a workflow. Start with one agent and split larger jobs when useful.

Build and check a project

Ask the agent to inspect a project, make a change, then run its tests. Follow the file edits and command results as it works.

Reuse work that repeats

Turn a repeated task into a script or saved workflow. Use codewhale exec in scripts and CI, or give parts of a larger job to several agents.

Stay in control

Set permissions before work starts, respond to approval requests, and interrupt a running task. Review the conversation and tool results before continuing.

Tools, connected apps, and saved work

Explore integrations

Files and commandsRead a project, edit files, run tests, and inspect command output within the permissions you set.
Plugins and MCPConnect additional tools and services. Review and enable plugins before the agent can use them.
Computer Use · previewA plugin that lets the agent see and operate other applications. Enable it explicitly and grant the system permissions it asks for.
Saved sessionsKeep the conversation and tool results together. The local browser connects to the same Codewhale session on your computer; resume saved work instead of starting over.
FleetAssign parts of a task to agents with different models and roles, and follow their progress.

A choice of models for every task

Choose the provider and model for each session: connect with an API key, use a supported provider sign-in, or run a local model. Your Codewhale account and your model connection serve different purposes.

Explore models and providers

Hosted: Your own API key, saved with codewhale auth set --provider <id>
Gateway: One endpoint for many models, provider still chosen by you
Local: vLLM, SGLang, Ollama on localhost — usually no key
Plan · Work · Operate: Ask · Auto-Review · Full Access

Getting started with Codewhale

Install the published release, connect a model, then try one task in your project folder. A team of agents is optional; start with one and add more when the work can be split.

1
Install Codewhale
The command below installs the latest published release on macOS or Linux. Use the install guide for Windows, package managers, or building the unreleased source candidate.
```
curl -fsSL https://codewhale.net/install.sh | sh
```
Full install guide
2
Connect your model
Codewhale needs a model to answer. Save your own provider key, as in the DeepSeek example below, or run a local model such as Ollama, which needs no key. You pay the provider directly.
```
codewhale auth set --provider deepseek
```
Connect a provider
3
Give it a task
Open Codewhale in your project folder and ask for something concrete. Start in /mode plan to have it explain the project without changing anything, then switch to /mode work for edits and commands. It shows each edit as a diff and asks before running a shell command.
```
codewhale
```
Set modes and approvals
4
Add a Fleet when you need one
When a task would benefit from several models and roles, run /fleet setup inside Codewhale to save roles and the model each one uses. From your shell, codewhale fleet status counts the Fleet runs that are queued, running, or finished.
```
/fleet setup
codewhale fleet status
```
Run a workflow

Read the getting-started guide Look up a term

Where you can use Codewhale

The terminal and local browser client are available now. Desktop and hosted web apps are being developed around the same session model; their availability is listed separately below.

Terminal and local browser: ReleasedInstall on Linux, macOS, or Windows. Run codewhale in your terminal, or codewhale web for the local browser client. npm and Cargo are alternatives; Android on Termux is a preview.
Hosted web app: Development previewSign in with a Codewhale account, then type /rc in a running terminal session to continue it from the web app. Hosted task execution is still being qualified.
Desktop: Development buildThe macOS app brings folders, conversations, and model connections into a desktop window. A public download is coming later.
Cloud computers: In developmentHosted computers for running your tasks.

The terminal and local browser do not require a Codewhale account. An account is used for hosted web and desktop access; it does not replace your model connection. Hosted model usage with your own key is billed by that provider.

Create an account

Install Codewhale on macOS or Linux

GitHub Releases (Binaries) · npm · Cargo · Docker · Windows · Android / Termux · China mirrors

Read the install guide

curl -fsSL https://codewhale.net/install.sh | sh

Help make Codewhale better

Whether you have found a bug, have an idea for a feature, or want to send your first pull request, we would like to hear from you and work together on what comes next.