AI assistants

Ramo can use an AI assistant you already have. It ships no model and runs no servers: requests go straight from your Mac to the assistant you connect. Everything here is optional and off until you connect one.

Connect an assistant

In Settings › AI assistants, connect one of:

• The Claude, ChatGPT (Codex) or Gemini command line tool, if you have it installed and signed in.
• Any OpenAI-compatible service with an API key, or a local model.

What it can do

• Suggest a commit message from your staged changes (⌥⌘G).
• Explain a commit in History.
• Suggest a branch name from your changes.
• Draft a pull request description (Repository menu).
• Suggest how to resolve a conflict (Beta).

Each one can be turned on or off in the same pane, and the instructions sent to the assistant can be changed in Settings › AI prompts.

What is sent

Only what the task needs, such as the staged diff for a commit message. With Ask before sending a diff on, Ramo shows the exact text first. Files matching Never send these files (for example .env and keys) are always left out, and a size limit keeps large diffs from being sent.