Introduction

TheGitAI is an interactive terminal coding agent. Local repo indexing, file edits, and shell commands run on your machine. Model inference and server-executed tools run on the server. You stay in the terminal you already live in, and the agent works directly against the repo you're in.

New here? Install the CLI, sign in once, and run ai inside any repo to start your first session. The next few pages walk through exactly that.

What you can do with it

In a single session, TheGitAI can:

  • Chat with you across multiple turns, remembering prior messages, tool outputs, and repo changes during that session.
  • Inspect files and directories, and read documents like PDF and XLSX.
  • Create, update, and delete files inside the repo you point it at.
  • Run non-interactive shell commands when needed (behind your approval).
  • Look up definitions, symbols, signatures, and diagnostics across your code.
  • Reindex changed files so later turns see your latest code.
  • Fetch public URLs when a task needs information from the web.
  • Migrate projects between frameworks, scaffold features, and validate changes with your build and test commands.

How a turn works

Each turn follows a simple loop:

  1. Your message is added to the live session history.
  2. The agent gathers the most relevant context from your repo.
  3. The current mode controls approval behavior and which tools are available.
  4. The model receives your request, the relevant context, and the available tools.
  5. The model either answers directly or calls a tool.
  6. The loop continues until the model is done.

Local file, search, and shell tools run on your machine; the model runs on the TheGitAI server. You approve anything that touches your files or shell.

Where to go next