> ## Documentation Index
> Fetch the complete documentation index at: https://ade-ac1c6011-ade-mobile-chat-stability-54bd7ef0.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Agent chat

> Talk to coding agents that can read, edit, test, commit, and help create PRs inside a lane.

Agent chat is the fastest way to get work done in ADE. Open a chat in a lane, describe the task, and watch the agent use files, shell commands, Git, tests, and PR actions.

## When to use chat

Use chat for:

* Small features and bug fixes.
* Codebase questions that need real file inspection.
* Test failures and CI failures.
* PR description and review follow-up.
* Iterating on an agent's first pass.

For broader coordination across lanes, use the CTO and workers.

## Providers

ADE can run chat through several backends depending on your setup:

| Provider             | Typical setup                         |
| -------------------- | ------------------------------------- |
| Claude / Claude Code | API key or authenticated CLI          |
| Codex                | Codex CLI or OpenAI-backed setup      |
| Cursor               | Cursor API key through the Cursor SDK |
| OpenCode             | Local OpenCode configuration          |

Switch provider and model from the chat header.

## Chat scope

Every chat has a scope:

| Scope              | What it means                                                              |
| ------------------ | -------------------------------------------------------------------------- |
| **Lane chat**      | Edits files and runs commands in one lane. This is the normal coding flow. |
| **CTO chat**       | Looks across the project and can delegate to workers.                      |
| **PR-linked chat** | Focuses on a PR, its CI, comments, and review tasks.                       |

## What you see

* Streaming agent responses.
* Expandable tool calls.
* Terminal output.
* File changes in Files.
* Questions when the agent needs input.
* Proof artifacts when the agent captures evidence.

<Tip>
  Keep the first prompt specific. "Fix the failing auth redirect test" works better than "clean up auth."
</Tip>

<CardGroup cols={2}>
  <Card title="Chat capabilities" icon="toolbox" href="/chat/capabilities">
    See what chat agents can do.
  </Card>

  <Card title="Context and artifacts" icon="database" href="/chat/context">
    Learn how context, transcripts, and attachments work.
  </Card>
</CardGroup>