Getting started

Go from a fresh installation to a project where Rem can help and a workflow you can run again.

1. Install Raticode

Open the download section and choose your desktop platform. The current published artifacts cover Windows x64, Linux x64, and macOS Apple Silicon. The desktop packages include the workflow backend; you do not need to install Python or Node.js to use a packaged app.

• Windows: run the setup executable. Leave Add the Raticode CLI to my user PATH enabled if you want terminal access. Open a new terminal after installation.
• macOS: open the disk image, copy Raticode into Applications, then open it. Follow the release's signing guidance if macOS blocks first launch.
• Linux: install the .deb or .rpm, or mark the AppImage executable and open it.

The installation guide has platform commands, CLI-only setup, and source development instructions.

2. Connect a provider

Install and sign in to one supported coding CLI using its own setup instructions. Raticode discovers local providers; it does not create a provider account for you.

Open Settings > Providers. Enable the provider you want to use and check its executable and model catalog. If it is not found, set its Executable override and choose Refresh providers. Then open Settings > Rem to choose the Default provider for new tasks and threads, and optionally a separate Rem chat provider for Rem's own replies.

Start with the provider you already use. See Provider setup for the CLI names and authentication checks. You can skip this step while learning command and file nodes.

3. Open a project

Choose Open Project (File > Open Project..., or Open project... in the project dropdown) and select a real project directory. The folder becomes the working context for files, source control, terminal sessions, and project-scoped Rem threads. Selecting a folder also gives the desktop permission to access it.

Check the project name before asking Rem to edit anything. A conversation keeps its own scope when you switch the editor to another project. See Projects and workspaces.

4. Send a first message

Open Rem, choose the provider and model in the composer, and review its permission mode. Open New thread tools, skills & MCP to inspect the resources the first turn will receive.

A useful first request is:

Read this project and explain its entry points, how to run it,
and where its tests live. Summarize your findings before changing files.

Press Enter to send. Shift+Enter adds a line break. Rem records the conversation in its thread; choosing another model later keeps that thread and its resource selections.

5. Create a workflow

Open Workflows and create one in the project, or ask Rem:

Create a Rattish workflow in this project that writes a short text report.
Validate it and check preflight. Do not execute it yet.

Follow Your first workflow for an example that writes two local files without calling a provider. Registered sources live in .raticode/<workflow-id>/workflow.rattish.

6. Validate, review, then run

Validation checks the language and graph. Preflight checks the machine's resources. Neither runs the workflow. Review any command, file write, agent prompt, and external request before choosing Run.

Open the bottom panel to inspect Problems, Run Timeline, and Runs. If a step fails, use its diagnostic rather than guessing from the final status.

Continue with Code and graph editors, Chat and threads, or Rattish language.