Guides
Set up Mandri for your projects and learn how to work with an agent in the app.
Get started
Section titled “Get started”- Run your first task: open a project, choose your agent’s settings and send a request.
- Add a model provider: connect an API account or a local model server from the app’s settings.
- Follow a task and review changes: read the agent’s activity, respond to approval requests and inspect changed files.
Model setups
Section titled “Model setups”-
Use OpenRouter: add your OpenRouter API key and choose a model for your task.
-
Use LM Studio: connect Mandri to a model served by LM Studio on your computer.
-
Use your ChatGPT subscription: connect your account and choose its model with Codex in the app.
Terminal
Section titled “Terminal”- Get started in the terminal: install the CLI, register a provider and launch an agent from your project folder.
- OpenRouter with Codex: configure your API key and select an OpenRouter model.
- LM Studio with Codex: connect a local model server.
- Use your ChatGPT subscription: sign in through the browser and launch Codex from the terminal.
Privacy
Section titled “Privacy”- Use free models with less exposure: configure pseudonymization, Docker execution or both in the app and terminal, and understand their limits.
- Privacy roadmap: planned controls for replacement values and Docker images.
If you have not installed the app yet, download Mandri for your computer.