AI Agents
Daintree's seventeen launchable built-in AI coding agents: the roster, the searchable launcher, the launch gate when a CLI is missing, alt-screen versus inline rendering, and permission bypass.
Built-in Agents
Daintree ships with seventeen launchable built-in agents. Each one is a third-party CLI that Daintree runs as a process in its own terminal panel with full PTY emulation. Daintree does not bundle the binaries and does not wrap the models: it launches the CLI you already installed, watches its output, and gives it a place in the habitat.
| Agent | Vendor | CLI | Status |
|---|---|---|---|
| Claude Code | Anthropic | claude | Supported |
| OpenCode | OpenCode (open source) | opencode | Supported |
| Aider | Aider (open source) | aider | Supported |
| Gemini CLI | gemini | Deprecated | |
| Antigravity | agy | Supported | |
| Codex CLI | OpenAI | codex | Supported |
| Grok | xAI | grok | Supported |
| Cursor | Cursor | cursor-agent | Supported |
| GitHub Copilot CLI | GitHub | copilot | Supported |
| Goose | Block | goose session | Supported |
| Amp | Sourcegraph | amp | Supported |
| Crush | Charm | crush | Supported |
| Qwen Code | Alibaba | qwen | Supported |
| Kimi Code | Moonshot AI | kimi | Supported |
| Open Interpreter | Open Interpreter | interpreter | Supported |
| Mistral Vibe | Mistral | vibe | Supported |
| Kiro CLI | Amazon | kiro-cli | Supported |
Antigravity (agy) is Google's consumer successor to Gemini CLI, and Daintree's integration checks ~/.gemini/ for its credentials. Grok is xAI's CLI.
Three agents ship default keyboard shortcuts: Claude Code (Option-Command-C Control-Alt-C Control-Alt-C ), Gemini (Option-Command-G Control-Alt-G Control-Alt-G ), and Codex (Option-Command-X Control-Alt-X Control-Alt-X ). Every other built-in agent has a launch action with no combo bound to it, ready for you to claim in Keyboard Shortcuts, from the agent's own page in Settings > CLI agents, or from the launcher itself. Most agents run on all three platforms. Crush and Kiro CLI ship install paths for macOS and Linux only, and Amp has no native Windows binary (npm or WSL).
Installing and Launching
Daintree does not bundle any agent. It detects the CLIs you have installed, gates the launch when one is missing, and starts the agent in the selected worktree. Installing each CLI, the searchable launcher, starting an agent from the new worktree dialog, alt-screen versus inline rendering and permission bypass are all covered at Installing and Launching Agents.
Presets
Presets are per-agent launch configurations: environment variables, custom arguments, a display title, a color, a fallback chain, and overrides for the bypass and screen-mode decisions above. Sources, per-worktree defaults, fallback chains, shared project presets, and provider templates are documented at Agent Presets.
Agent States
Daintree reads each agent's PTY output and tracks what it is doing (idle, working, waiting, directing, completed, or exited) along with why a waiting agent is waiting. Those states drive the panel header, worktree cards, the sidebar filters, Fleet, Pilot, and notifications. The full reference is at Agent States.
Claude Code and Codex can also delegate to subagents of their own. When one does, a chip in that terminal's header lists the child sessions and the task and state information the provider can report. See Subagents in the Header.
Launch Hint vs Runtime Identity
Daintree separates what you asked to launch from what is actually running. The distinction earns its keep whenever a session pivots in place: an agent exits to a shell prompt, you type claude, and a different agent starts in the same panel.
The launch hint (launchAgentId) is persisted with the panel. It drives command generation, restart, resume flags, and preset and model lookup. Current chrome follows runtime identity; the launch hint remains available to restoration and diagnostic controls.
The runtime identity (detectedAgentId) is not persisted. It is resolved live from the process tree, by the executable's image path rather than its window title or argv (CLIs rewrite both), and it drives the chrome: icon, color, panel border, dock fade, focus navigation, and fleet-arming eligibility.
A plain terminal is promoted to an agent panel the moment you type a registered agent command, through the same code path a launcher launch uses. When the agent exits back to the shell, the panel demotes to a generic terminal. Demotion keeps the persisted launch hint; conversation recovery still depends on a usable session ID and the CLI's resume integration.
Session Resume
When an agent supports resume and Daintree captures its session ID, closing the panel, letting its trash window expire, or quitting records a resume entry. The launcher offers that entry back with the CLI's resume flag. The conversation itself remains in the CLI's own storage, so recording an entry is not a transcript backup.
Claude Code session IDs are assigned at launch, which makes them available for recovery after a crash. Most other session-ID integrations capture the ID from a clean shutdown; a force quit can lose that capture. Cursor and Crush have no resume integration. Aider, Open Interpreter and Kimi Code expose rolling-history continuation, while Kiro resumes by directory, but Daintree's current automatic restart and restore paths do not invoke those ID-less modes. They start fresh unless you supplied continuation arguments yourself. Use the CLI's documented manual continuation command when you need its earlier conversation. The per-agent syntax is on each agent's page; the journal, retention, bookmarks, and the trash window are documented at Resuming a Closed Agent Conversation.
An agent Daintree launched can also quit back to its shell while the panel stays open: a crash at boot, a quit, or a cancelled update. When that run leaves no resume hint behind, the pane says so with a banner, <Agent> quit to the shell, and offers Resume conversation to reopen the conversation the pane was opened on. A Codex launch with no recorded session offers Codex's Find session picker instead. Codex's own self-update is handled for you: when Update now succeeds and Codex exits asking to be restarted, Daintree relaunches the same conversation in that pane, once.
Moving a project
A project folder can be moved or renamed from inside Daintree, and whether an agent's conversation survives that is a property of how its CLI stores history. Daintree classifies each agent before the move and shows the verdict in the relocation preview, so you find out beforehand rather than on the next resume:
- Preserved: the session id is globally addressable, so it resumes at the new path (Codex, GitHub Copilot).
- Travels with the project: the history file lives inside the folder and moves with it (Aider).
- Needs a manual step: the conversation still exists on disk, but the provider cannot resume it at a new path. Claude Code's store is keyed by a path slug; Kiro's by an absolute path. Daintree will not reach into a provider's private store to fix that.
- Unavailable: no usable resume path across a move (Gemini CLI, Crush).
An agent Daintree has not classified is reported as unverified rather than assumed safe.
Models and Context Windows
Model choices start with the bundled agent registry. Daintree enriches supported providers from models.dev, but curated agents restrict the offered IDs. Claude offers its curated aliases (Opus, Fable, Haiku, Sonnet). Codex offers its curated list (GPT-6 Astra, GPT-6 Sol, GPT-6 Luna), and the installed CLI's codex debug models --bundled catalog only prunes the IDs that CLI does not know; if it knows none of them, the older CLI's own list is offered instead. A remote catalog entry does not by itself guarantee that a CLI accepts that model. Context-window numbers are metadata estimates; the selected model and provider determine the actual limit.
Plugin-Contributed and Custom Agents
The built-in roster is not the whole list. A plugin can register agents of its own through contributes.agents, gated by the agent:register capability. A plugin-contributed agent is a first-class agent: it appears in the launcher, launches into a worktree, takes presets, and drives the working/waiting/completed state UI. A plugin may also declare output-detection patterns for its agent, which sharpen prompt and completion detection beyond what output volume alone can tell.
Plugin agents merge into the effective registry at the lowest priority. They are purely additive for new ids and can never shadow a built-in, so installing a plugin cannot quietly change what claude means.
Daintree also carries a user registry for custom agents: an id, a display name, a command and args, a color, an icon, a shortcut, and whether the agent accepts injected context. Custom agents sit above plugin agents and below built-ins in the same merge, and appear alongside them in every launch surface.
Completions
Daintree's unified input can complete an agent's own commands as you type. What it offers is declared per agent in the registry rather than guessed, so each CLI gets its real vocabulary: Claude Code, Codex, and Gemini declare completion sources today.
Codex is the richest of the three. Its / trigger lists built-in slash commands; its $ trigger surfaces Codex Skills and Plugins together, each row badged with its category. Skills come from Codex's system and user skill folders under CODEX_HOME, your personal ~/.agents/skills (which wins a name clash with CODEX_HOME's), and the project's .agents/skills; symlinked skill folders are followed. Plugins come from the enabled entries in your Codex plugin registry. A skill you wrote wins a name collision against a plugin of the same name. Codex Apps resolve server-side, so they do not appear in local discovery.
Quitting While Agents Work
If any agent is in the working state, or is waiting on a permission prompt, when you quit Daintree, the quit is intercepted and a dialog titled "Agents are working" reports how many of each. Cancel is the default. With a window open, Keep Agents Running closes every window and leaves the agents running in the background until you open Daintree again, and Quit and Stop Agents proceeds with a graceful shutdown; with no window open, the choice is Quit Anyway. See Keeping agents running when you quit. Agents waiting for any other reason, directing, idle, or finished do not trigger it. These classifications are observations of terminal activity; check the terminals before quitting if a long-running job appears quiet. OS-level signals such as SIGTERM and a force quit skip the dialog entirely.
While at least one agent is working, Daintree can also hold off idle sleep, so power management does not interrupt a long-running task. The display can still turn off. This is Settings > General > Overview > Keep awake: Keep awake while agents work is on by default, and Keep awake on battery is off by default, so on a laptop the hold applies only while it is plugged in. Plugging in and unplugging are honored live. The hold lifts as soon as the last agent stops working, and a single run of work can hold the machine awake for at most twelve hours (checked every four), so a wedged run cannot keep it awake indefinitely.
Broadcasting to Many Agents
To drive several agents at once (arming terminals, previewing what a broadcast will send to each worktree, and running the destructive fan-out actions across an armed set) see Fleet. There is no separate bulk-operations palette; those actions live in Fleet.
Handing a Terminal to an Orchestrator
An agent pane connected to Daintree's MCP server can supervise the agents it launches. You can also give it one it did not launch. Right-click the terminal and choose Hand to orchestrator, then pick the pane. The submenu lists only panes in this view that hold a live Daintree MCP connection allowed to drive terminals, and it does not appear when there are none.
A confirmation names both panes and says what the orchestrator gets: it can type into the terminal and submit, interrupt it, bring it on screen and read its last message, as if it had launched it. It cannot close the terminal. While the hand-over lasts, the terminal's header carries a Driven by chip naming the orchestrator. To end it, right-click the terminal again and choose Take back from the orchestrator (or Take back control); the hand-over also ends when the terminal's process exits.
1 can approve a push. Hand over a terminal only to a pane you would trust at its keyboard.A hand-over is refused when the orchestrator launched the terminal itself (it already drives it), when the terminal is already handed to another pane (take it back first), or when the orchestrator belongs to another project.
Sending Context to an Agent
Use CopyTree to hand an agent a structured snapshot of the codebase: Shift-Command-C Control-Shift-C Control-Shift-C copies one to the clipboard, and the Copy Context control on a worktree card offers full context or modified files only, backed by per-project history.
Send selection to another panel
To move text between panels without the clipboard, select it in any terminal or agent panel and press Shift-Command-E Control-Shift-E Control-Shift-E (or right-click and choose Send to agent). A searchable list of the other PTY-backed panels appears, each naming its worktree when they span more than one; the text lands in the target's input buffer as if typed, unsubmitted, so you press Enter yourself. Panels whose input is locked are listed but not selectable.
Daintree Assistant
The Daintree Assistant is Claude Code, Codex or Antigravity, launched in the assistant sidebar with Daintree's own docs, runbooks and a tool set wired in. Open it with Command-L Control-L Control-L or from the toolbar. It answers questions about Daintree and, within the tool set you choose, drives the app for you: creating worktrees and launching, prompting and closing agents. It runs on your own subscription, with its own agent, model and permission settings, separate from everything above. What it can do and how it asks first are documented at Daintree Assistant.
MCP Server
Everything above is about agents running inside Daintree. The MCP server inverts that: Daintree becomes the server, and external clients connect to it from outside to read and act on your habitat. Setup, the tool tiers, authorization, and the audit log are documented at MCP Server.