Skip to main content
Screenshot: Agent Chat pane in a Lane view showing the chat composer, a conversation with tool call blocks, the context meter, and the MCP tool indicator.

​
What is Agent Chat?

Agent Chat is ADE’s conversational interface for working with AI agents in the context of your actual project. Unlike a generic AI chat, Agent Chat is wired directly into your development environment: the agent can read and write real files, run terminal commands, create branches and PRs, query your project’s context pack, and spawn a Mission when the task grows beyond what a single conversation can handle. Every chat session is recorded with a full transcript, tool call log, and timing data. Sessions are resumable and their artifacts can be promoted into the lane’s persistent context.

What Chat is for

Exploratory work, targeted edits, quick Q&A about your codebase, PR creation, test runs, and anything that resolves in a single focused session.

When to escalate to a Mission

Tasks spanning multiple files or agents, work requiring a formal audit trail, or anything you expect to run for more than 15–20 minutes.

​
Chat Scoping

Chat in ADE is always scoped — it operates within a defined context boundary that determines what the agent can access and what its tool calls affect.
Lane chat is the most common mode. It is opened from within a specific Lane view and anchors the agent to that lane’s context:
  • File reads and writes are scoped to the lane’s git worktree
  • Terminal commands run in the lane’s terminal session
  • Git operations affect the lane’s branch
  • Context is seeded from the lane’s pack (recent activity, file map, open PRs)
Lane chat is how you do day-to-day development work with an agent.

​
Starting a Chat

1

Open a Lane

Navigate to the lane you want to work in from the left sidebar or the Workspace Graph.
2

Open the Chat pane

Click the Chat button in the lane toolbar, or press Cmd+K. A new chat session opens with automatic context injection from the lane’s pack.
3

Choose a model (optional)

The model selector in the chat header defaults to your configured preferred model. Click it to switch models for this session. ADE remembers your last-used model per lane.
4

Set execution mode (optional)

For CLI-wrapped models (Claude, Codex), an execution mode selector appears in the header:
  • Focused — single-threaded; the agent works through the task in one thread
  • Parallel — the agent can spawn parallel sub-agents for independent parts of the task
Use Parallel mode for tasks the model can naturally decompose (e.g., “add tests for all these functions”).
5

Type your first message

The agent receives your message along with the injected context. Tool calls begin immediately if the task requires them.You can also attach images (JPEG, PNG, GIF, or WebP) to your message using the attachment button in the composer. Images are sent inline as base64 content blocks, so the agent sees them natively — no file-read workaround required.
Start a new session from an existing session’s history by opening the History tab and clicking Resume. The previous transcript is loaded as context so the agent picks up where you left off.

​
Computer Use Policy

Screenshot: Chat header close-up showing the computer-use policy chip, MCP tool count badge, model selector, and context meter.
The chat header displays the current computer-use policy for the session. This controls whether and how the agent can invoke computer-use capabilities (browser automation, screenshot capture, desktop interaction). Click the policy chip to change it for the current session.
The agent cannot invoke any computer-use tools. Use this for sessions focused purely on code changes where you do not want any visual verification steps.
ADE enables computer use when the agent’s task naturally calls for it — e.g., verifying a UI change, testing a web flow, or checking a rendered output. The agent will not use computer use for tasks that can be completed via file and terminal tools alone.
Computer use tools are always in the agent’s tool palette. The agent may invoke them freely. Use this when you are explicitly working on UI-heavy tasks.
The agent first attempts to complete the task without computer use. If it cannot (e.g., a required verification only works via a browser), it falls back to computer use automatically.
Computer use is available, but every action must produce a proof artifact (a screenshot or browser trace). All proof artifacts are collected and can be attached to a PR or Linear issue. Recommended when the session will feed into a code review.
When computer use is active, a monitoring panel expands below the chat. It shows:
  • Recent screenshots captured during the session
  • Browser action log (clickable to expand each action)
  • A proof summary for the current session

​
Explore More

Chat Capabilities

What the agent can do — file operations, git, PRs, missions, MCP tools, and more.

Context & Artifacts

Managing context windows, working with artifacts, MCP tool visibility, chat history, and troubleshooting.