> ## Documentation Index
> Fetch the complete documentation index at: https://docs.v3code.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Modes

> Ask, Edit, Plan, and Agent — one agent, four ways to work.

V3Code has a single agent with four modes. The mode changes what the agent is allowed to do with a turn — everything else (your model, your keys, your context) stays the same. Switch modes from the picker in the chat input.

## The four modes

| Mode      | What it does                                                                                               | Reach for it when                                                                    |
| --------- | ---------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------ |
| **Ask**   | Pure conversation about your codebase. No tools, no edits — just answers.                                  | You want a quick explanation without the agent touching anything.                    |
| **Edit**  | Read-only investigation with the full tool set — reads, searches, structural lookups — but no file writes. | You want the agent to dig in and report back before changing code.                   |
| **Plan**  | Lays out an approach and can write **Markdown only** (like a `PLAN.md`); no code edits.                    | You want an agreed plan before execution. Research is budgeted so it doesn't spiral. |
| **Agent** | Full execution — reads, edits, runs commands, the whole tool set.                                          | You're ready to ship a change end to end.                                            |

Each mode pairs with whatever model and reasoning effort you've picked, so you can run **Plan** on a strong model for the thinking and **Agent** on a fast one for execution.

## Approvals: what needs your OK

Modes control *what class* of action the agent takes; **approvals** control whether a given action runs automatically. Reads are automatic. Actions that change things ask first, grouped into a few classes:

* **Edits** — creating, deleting, rewriting, or editing files, renaming symbols, generating images.
* **Terminal** — running commands, the test runner, the sandbox, and any git write (commit, push, checkout…).
* **Browser & notes** — navigating/typing/clicking in the browser, and saving persistent notes.

Read-only tools (reading files, searching, structural lookups, git status/diff) never prompt. If you want the agent to stop asking, **YOLO mode** auto-approves everything — behind a clear warning, since it applies everywhere.

## Modes vs. the V3 / IDE layout

Don't confuse the agent's modes with the **V3 / IDE** pill in the title bar — that switches the *layout* (full-screen chat vs. classic editor), not the agent's behavior.

<Note>
  "Design" isn't a mode — it's an [agent skill](/editor/design-mode) you invoke to build on-brand UI.
</Note>