The problem: writing in one window, agents in another
Most people who work with AI writing assistants end up with a split screen problem. The editor is in one window. The AI agent runs in a separate terminal. You write something, switch to the terminal, paste a prompt, wait for output, copy the result, switch back to the editor, paste it in. Repeat.
It works, technically. But it breaks your focus every time you switch. And the agent has no idea what your document looks like unless you manually give it context.
How Ritemark solves this
Ritemark has a full terminal built into the editor. Not a simplified command line, not a chat widget pretending to be a terminal. An actual xterm.js terminal panel that runs whatever CLI tools you have installed.
Open the terminal panel on the right side of your editor. If you have installed Claude Code, run claude to start it. Run gemini to start Gemini CLI. Run any other agent that ships as a command-line tool. The agent starts up inside Ritemark and can read and edit the same files you have open in the editor.
You write on the left. The agent works on the right. Same files, same project folder, same session.
What this looks like in practice
Say you are writing a technical guide and want the AI agent to review your draft for accuracy.
You have the document open in the visual editor. You open the terminal panel and start Claude Code. You type: "Review the document I have open and flag anything technically inaccurate." The agent reads the file directly from your project folder, reviews the content, and suggests changes. If you approve, it edits the file in place. The editor updates with the changes.
No copying. No pasting between apps. No "here is the context" preamble every time.
Or maybe you want to generate a section outline. You tell the agent: "Read the first three sections and propose an outline for section four that matches the existing tone." The agent reads your actual Markdown files, understands the structure, and generates an outline that fits.
Not locked to one agent
The terminal is agent-agnostic. Claude Code, Gemini CLI, OpenAI Codex, or any future CLI-based AI tool. If it runs in a terminal, it runs in Ritemark.
You can also switch between agents depending on the task. Use Claude Code for long-form editing work. Try Gemini CLI for quick research questions. If you install Ollama yourself, run a local model for sensitive content. The AI sidebar has no local model option. The terminal does not care which agent you choose.
The AI Sidebar alternative
Not everyone wants a terminal, and you do not need one. Ritemark's AI Sidebar is a visual chat interface with Claude Code, Codex and OpenCode built in, so there is nothing to install. You sign in with your paid Claude plan or ChatGPT account, or add your own API keys. In the sidebar you can switch between agents, choose Manual, Auto or Plan only (Plan only for Claude Code and Codex), and answer the agent's questions.
Both interfaces work on the same project files. The terminal is for people comfortable with the command line. The sidebar is for everyone else. You can use both in the same session.
Your files, your machine
The terminal runs locally. Your project files stay on your hard drive. The agents make API calls to their own providers (Anthropic, Google, OpenAI), and the AI sidebar sends what the agent reads, such as your open file, selections and attachments, to the provider you pick. There is no Ritemark cloud, no account, no sync service. Anonymous usage analytics are on by default and can be switched off in Settings.
Download Ritemark and start working with AI agents directly in your writing environment.
Start writing with Ritemark.
Free and open source.