Skip to main content

Global Settings

Daintree's 24 app-wide settings tabs across the General, Terminal, Assistant, Integrations, and Support groups, including skip permissions, alt-screen mode, plugins, the MCP server and Hosts.

Updated
Reviewed

Global settings cover 24 tabs across five sidebar groups: General, Terminal, Assistant, Integrations, and Support. They apply to Daintree as a whole, regardless of which project is open. For the per-project layer that overrides them, see Project Settings.

This page is a map, not a field-by-field mirror of the dialog. Each tab gets a line on what you'd come here to change, and where a subsystem has its own page, the link goes there instead of repeating it.

Filtering needs JavaScript. The contents list works without it, and so does your browser's find-in-page — every tab on this page is in the document.

The tabs

GroupTabCome here to change
GeneralGeneralAgent status, restoring live projects, where opened folders go, keeping the machine awake, update channel, hibernation and idle auto-close, display indicators
GeneralAppearanceApp theme, color vision mode, dock density, terminal color scheme and font
GeneralKeyboardRebind shortcuts, import and export shortcut profiles
GeneralNotificationsWhich agent events notify you, and what they sound like
GeneralLinksYour own links, offered in the links launcher in every project
GeneralPrivacy & dataTelemetry level and its live preview, log and session-history retention, data folder, factory reset
GeneralImport & exportExport your configuration to a file, or import one from another machine
TerminalPanel gridPanel limits, scrollback, split layout, cached project views, screen reader mode
TerminalWorktreeWhere new worktrees are created, what happens to a deleted worktree's terminals, and which files the file browser always hides
TerminalToolbarWhich buttons sit on which side of the toolbar, and launcher defaults
TerminalEnvironmentEnvironment variables injected into every terminal Daintree spawns
AssistantDaintree AssistantWhich CLI runs the assistant, how much it's allowed to do, and what gets logged
IntegrationsCLI agentsPer-agent flags, presets, skip permissions, alt-screen mode, install help
IntegrationsCode forgeForge credentials, default provider, and how each remote resolves
IntegrationsIntegrationsExternal editor and image viewer
IntegrationsVoice inputTranscription provider and key, recording mode, dictionary, AI correction
IntegrationsHostsMachines this one opens projects on over SSH (WSL distributions on Windows), cloud provider accounts, and whether this machine can be a host (macOS and Linux)
IntegrationsMCP serverTurn the local MCP server on, copy a client config, see who's connected
IntegrationsPluginsOpen the plugin manager; turn background update checks on
IntegrationsPlugin actionsRead (or stop recording) the audit log of actions plugins dispatched
IntegrationsCanopyShow or hide Canopy, the agent inbox, and whether it reads agent terminals
IntegrationsDecisionsAnswer agents' questions from your phone through a relay in your own Cloudflare account
IntegrationsRun historyReview past recipe runs and fleet broadcasts
SupportTroubleshootingHardware acceleration, health check, diagnostics report, logs and log levels, developer mode
The lower sidebar holds the Integrations and Support groups

General group

General

Three subtabs: Overview, Hibernation, and Display.

Overview, from the top:

  • System status summarizes the agent CLIs it found (for example, all installed agents ready). An agent that needs you is listed with what is wrong: Needs setup, No credentials detected, or Blocked. Ready agents sit behind a Show N ready agents disclosure, and a Daintree supports N more agents link opens CLI agents for the rest of the roster. Clicking an agent row jumps straight to that agent's page. With no agent found, it offers Run setup wizard.
  • Startup > Restore live projects (on) brings back every project that was running after a restart, not just the one each window was showing.
  • Opening folders > Open folders in a new window decides where a folder you open goes: Only from outside Daintree (the default: a folder opened from your file manager or the command line gets a new window, one picked inside Daintree uses the current one), Always, or Never. Switching between projects isn't affected.
  • Keep awake holds off idle sleep while an agent is working, and says whether it is doing so right now. Keep awake while agents work is on by default; the display can still turn off. Keep awake on battery, indented under it, is off by default, so an unplugged laptop is allowed to sleep. Plugging in or unplugging takes effect as it happens.
  • Updates > Update channel is Stable or Nightly, with a line saying when Daintree last checked. On a Microsoft Store build the section instead reads Updates are managed by the Microsoft Store on Windows and offers Notify when a new version is available. See Updates for how releases are delivered.
  • Quick reference is a collapsible card of common shortcuts, with a link to the Keyboard tab.

The Daintree version, architecture and build label sit in a strip at the foot of Overview, with a link to daintree.org.

Note
Nightly builds carry the latest work and correspondingly more risk; selecting Nightly changes the description under the control to say so, and you can switch back at any time. If the channel can't be read from disk the control stays blank, with a Couldn't load the update channel banner and a Retry, rather than falling back to "Stable". A nightly user seeing "Stable" selected as though it were authoritative is worse than an empty control.

Hibernation holds the Background projects controls. None of them touches the active project:

  • Notify me about idle terminals (on) raises a reminder once the terminals in a background project have gone quiet for the Remind after time (30 minutes, 1 hour by default, 2 hours or 4 hours), offering to close them. Nothing closes on its own. Covered in Notifications & Sound.
  • Close idle projects automatically (off) frees memory from background projects with no open terminals once they have sat idle for the Close after time (15 minutes by default, 30 minutes, 1 hour or 2 hours). They stay in the switcher and reopen with their panels. This runs separately from the disabled terminal hibernation below; a running Daintree Assistant also prevents automatic closing.
  • Hibernate inactive projects (off) and Hibernate after remain visible, but automatic terminal and dev-server teardown is disabled in this candidate build. Use Free memory or Sleep idle projects for explicit reclamation. See Project memory.

Display holds six Interface elements toggles:

  • Project pulse (on) shows an activity heatmap on the empty panel grid
  • Developer tools (off) shows the Problems button in the toolbar; it also appears on its own while file watching is degraded
  • Grid panel agent highlights (off) shows waiting and working borders on grid panels
  • Dock item agent highlights (off) shows waiting borders on dock items
  • Agent task in terminal titles shows the agent's current task next to its name in tabs and panel headers
  • Reduce UI animations minimizes motion across the interface, independent of your OS reduce-motion setting

Failed-state borders are always drawn regardless of either highlight toggle. A failure is not decoration.

Appearance

Two subtabs. App holds App theme (the theme picker with 15 built-in themes, eight dark and seven light; Match system appearance with a dark and a light theme; an Accent color override that swaps the active theme's accent without editing the theme, with a low-contrast warning; and Theme files to load or save a theme) and Interface (Color vision: Default, Red-green or Blue-yellow, with a live preview; and Dock density: Compact, Normal or Comfortable). Terminal holds the terminal Color scheme (hover or focus one to preview it in your open terminals, or import one) and Font: family (bundled JetBrains Mono, or the system monospace font) and size (8–24px).

See Theme System for the theme list, auto-switching, authoring, and sharing.

Keyboard

View, search, and rebind every shortcut. Click a binding to record a new one; conflict detection warns when the combination is already taken. Shortcut profiles import and export as JSON, and shortcuts reset one at a time or all at once.

See Keyboard Shortcuts for the full reference and the profile system.

A conflict warns rather than blocks; Unbind clears the other action's binding before you save.

Notifications

Which agent events raise a notification, which sounds they use, and the quiet-hours window. See Notifications & Sound for the whole system, including per-project overrides.

One section, Your links: links offered in every project on this machine, shown in the toolbar's links launcher. Add a link with a name and an http:// or https:// address, edit it, move it up or down, switch it off to hide it without removing it, or remove it with Undo. See Links Launcher.

Privacy & data

Two subtabs: Telemetry and Data & storage.

Telemetry holds Telemetry & diagnostics, with three levels and a What's collected at each level disclosure listing exactly what each one sends:

  • Off: nothing is sent, and crash reports aren't submitted
  • Errors only: crash reports and error details, no usage analytics
  • Full usage: crash reports plus anonymous usage analytics. Analytics recorded before you chose a level may be sent too.

Preview outbound telemetry lets you inspect every sanitized payload Daintree would send, live and for the current session only; the preview itself transmits nothing. Crash reporting is configured here rather than under Troubleshooting, because it is a privacy decision before it is a diagnostic one. At Errors only, crash reports go to Sentry with home-directory paths redacted from stack frames and messages; the payload is the exception type and message, sanitized stack frames, app and Node versions, OS name and architecture, standard Electron SDK runtime metadata, and main-process breadcrumbs preceding the crash. Full usage adds a fixed set of anonymous onboarding and activation events, each carrying a name, timestamp, and event-specific properties. Known secrets and paths are scrubbed before upload, and native minidump uploads are disabled. Pattern-based redaction can miss arbitrary content in error details, so inspect the preview before sharing diagnostic data.

Note
A level change applies straight away. Choosing Off stops uploads mid-session, including reports already queued for retry; raising the level starts reporting without a restart.

For sampling rates and the pre-consent buffering rules (how events queued during first launch are handled once you pick a level), see Telemetry & Privacy.

Data & storage holds the destructive and semi-destructive controls, in three sections:

  • Local data: Data folder shows the path to Daintree's data directory with a button to open it; Log retention is 7 days, 30 (default), 90, or keep forever, pruned at startup so a change takes effect on next launch; Clear cache clears the HTTP disk and code caches for the app, browser panels and dev previews, leaving sign-ins, site data and settings alone; and Hidden commands restores commands you've hidden from Recently used in the command palette.
  • Session history: Keep session history for sets how long resumable agent-session records are kept, on the same 7/30/90/forever scale, and Clear session history deletes every recorded session now. Both apply to every project; shortening the window deletes older records straight away, so it asks first. Bookmarked sessions survive a clear. Unbookmark a session before clearing it.
  • Factory reset: Reset all app data deletes every setting, API key, recorded session and log on this machine, then restarts Daintree with factory defaults. Your repositories and their .daintree folders aren't touched.
Note
There are two clear-session-history controls with deliberately different reach. The one here clears the journal across all projects. The one on a worktree card clears only that worktree's records. See Clearing session history.
Warning
Reset all app data is irreversible. Daintree asks you to type Daintree to confirm, then restarts with factory defaults.

Import & export

One section, Configuration file, with Export configuration (Export…) and Import configuration (Import…). It replaced the old Help menu items in 0.38; both actions are also in the command palette. What travels, what is left out, and how the import preview and its backup work are covered under Move configuration to another machine.

Terminal group

Panel grid

Five subtabs: Performance, Input, Layout, Scrollback, and Accessibility.

Performance:

  • Performance mode cuts scrollback to a fixed low value and disables animations
  • Resource monitoring shows per-terminal CPU and memory in panel headers, with process-tree polling that adapts to the resource profile
  • Memory leak detection warns when a terminal's memory grows continuously, and can auto-restart it past a configurable threshold. Requires resource monitoring.
  • Panel limits sets how many panels you can open at once. Defaults are auto-detected from RAM and CPU cores, and Reset to hardware-recommended defaults puts them back. Three levels: Soft warning, Confirmation required past a batch spawn, and a Hard limit that can't be bypassed. Panel warnings turns the first two off; the hard limit always applies.
  • Project views > Cached project views (1–5) sets how many projects stay in memory for instant switching. The default is picked from system RAM on first launch, and the adaptive resource profile can clamp it lower under pressure, so the number you set is a ceiling, not a guarantee. See Cached Project Views and Performance.
Note
Performance mode applies to terminals as they are spawned. Panel limits govern new panel-creation attempts; raising or lowering a limit does not close existing panels.

Input holds Agent input: Hybrid input bar (on) shows the multi-line input bar at the bottom of agent terminals, and Focus the input bar first (on) decides whether selecting an agent panel starts focus in the input bar or in the terminal. Clicking either one still wins, and Option-Command Control-Alt Control-Alt +arrow follows whichever you're using. Once you use Remote Hosts, it also offers Send my clipboard images to remote agents on Ctrl+V (on), so an agent on a host gets the image on this computer's clipboard.

Layout holds Two-pane split (Split two panels with a divider, on by default, with a Preview-focused layout 65/35 variant, a Default ratio slider, and Reset all worktree split ratios) and Grid layout > Strategy for three or more panels: Automatic (default), Fixed columns, or Fixed rows.

Scrollback sets the base history buffer: 500, 1,000 (default), 2,500, or 5,000 lines. Agent terminals scale to 10× that base, capped at a 10,000-line policy maximum (raised from 5,000 in v0.28) and lowered further under the Efficiency resource profile; shell and dev-server terminals scale to 0.3× the base instead, capped at 2,000. An expandable estimate shows memory per terminal type. See Terminals for the full multiplier table and how the ceiling behaves under memory pressure.

Accessibility holds Screen reader mode: Auto (default, follows the OS), On, or Off. Turning it on adds an accessible DOM overlay to terminals, which costs performance, hence following the OS rather than defaulting to always-on.

Worktree

Three sections. Path pattern sets the directory template for new worktrees using the variables {base-folder}, {branch-slug}, {repo-name}, and {parent-dir}, with three presets (Grouped by repository, the default; Repository + branch; and Branch only) and a live preview of the resolved path. A relative pattern resolves from the repository root. Deleted worktrees > Close leftover terminals decides what happens when a worktree is deleted while its terminals still run: they wait in a temporary sidebar row and move to trash after 30 seconds, 1 minute (default), 5 minutes, or never. Always-hidden files lists names that stay hidden in every file browser panel, so generated noise doesn't have to be re-hidden per project.

See Worktrees for the pattern variables and File Browser for the file browser.

Toolbar

The toolbar has a left side and a right side, and this tab is where you decide what lives on each. Toolbar buttons lists the Left side and Right side: drag to reorder within a side, move a button across, or hide it. Not on the toolbar lists what you can add, grouped as Agents, Panels, Plugin buttons and Other buttons. Every agent already lives in the launcher, so pinning one to the toolbar is an extra, not a requirement. The Launcher palette section sets the Default selection and Always show dev server in launcher, even when no command is configured. Reset toolbar restores the default layout.

See UI Layout.

Environment

One section, Global variables: environment variables injected into every new terminal Daintree spawns, in every project. A project's own Variables override global values of the same name. Agent, preset, recipe and caller-specific launch environments can override these again.

The editor is a key/value list with explicit Save and Discard: changes don't apply until saved, and then only to terminals spawned afterward. An invalid key is flagged inline and raises the validation mark on the sidebar tab until it clears. If saved variables can't be read from disk, the tab disables editing rather than showing an empty form you could "save" over your real data, and says so.

For how this tab, the project Variables tab, and the per-agent env block divide the work, see Environment variables across scopes.

Assistant group

Daintree Assistant

The assistant runs Claude Code, Codex or Antigravity in the assistant sidebar, on your own sign-in and plan, and talks to Daintree through the local MCP server. This tab decides which CLI and model, which servers it connects to, how much of Daintree it can reach, and what gets recorded, for the project assistants and the global Assistant alike. Changes apply to new assistant sessions. Sections:

  • Agent: Agent picks the default CLI (Claude, Antigravity or Codex; the first start asks if more than one is installed), and picking one also switches this project's selected session to it. Model picks its model. Until you choose a model, the Assistant uses the CLI's own default; Default (CLI default) passes no --model flag. Notices here flag an agent that is no longer available or a CLI that needs an update
  • Name and emoji: a name and emoji of your own for this project's assistant and for the global Assistant
  • Behavior: Search documentation (connect the daintree-docs server), Daintree control (connect the local daintree server on 127.0.0.1, with its status underneath), and Follow runbooks (connect daintree-runbooks so the assistant looks up Daintree's procedure before acting on a task; needs Daintree control). All three are on by default
  • Security: the Tool set, the selected CLI's permission bypass, Daintree confirmations, a What this tool set allows disclosure, Let other machines' assistants work here (on by default, on macOS and Linux), and the Live session card
  • Privacy: Capture audit log, Audit log retention (7 days by default, 30 days, or Off), and an Advanced diagnostics disclosure with the audit-log viewer, latency table and turn-outcome diagnostics
  • Launch options: Custom CLI args appended to the launch command (a --model here overrides the picker) and Hibernate after, how long the panel stays hidden before the assistant shuts down (5 minutes by default; reopening resumes the same conversation)
  • Custom commands and skills: Assistant folder opens ~/.daintree/assistant, whose commands, skills, instructions and reference files join new sessions; a project's own .daintree/assistant takes precedence. Load my MCP servers and hooks (off by default) also loads that global folder's mcp.json (not for Antigravity) and Claude Code hooks.json. See Assistant Commands & Skills
  • External clients: Copy MCP config to point another MCP client at the same local server, and Rotate MCP key…, which disconnects every client using the old key

Tool set and the live session card

The Tool set is the set of Daintree actions a new assistant session can call. There are two:

Tool setWhat it allows
Core (default)Create worktrees, launch, prompt and resume agents, read and wait on their terminals, move, rename or close terminals, read issues, pull requests and CI, run a shell command, and reach the project's links and your remote hosts. It can delete any eligible worktree, with Daintree confirmation according to the policy below.
FullAdds recipes and workflows, project checks, git activity, CopyTree context, managing worktree resources, opening, killing and restarting terminals, and diagnostics. Deletions and teardowns still ask you to confirm.

Git and forge writes and file edits are in neither set; the Assistant does that work with its CLI's own tools, and files an issue with the forge CLI only after you approve its repository, title and body. A tier saved before the split reads in place: Workbench and Action as Core, System as Full. The What this tool set allows disclosure lists every action in the selected set by namespace, with the count (and, for Full, how many more than Core), and pins the high blast radius deletes, teardown and shell command at the top. The Live session card reports what the running session is actually doing: connected or not, the set it is running, and whether that sits above, below, or level with the configured default. A session can sit above the default because you approved Set project default on a tool-not-permitted banner. The card lists time-bounded per-tool grants with a countdown, and holds automation grants: approve a list of tool ids for a maximum number of uses, then watch or revoke them.

Warning
Bypass Claude permission prompts passes --dangerously-skip-permissions to the assistant's Claude session, skipping the per-tool confirmation gate for every tool, built-in and MCP alike. Bypass Codex approvals and sandbox passes --dangerously-bypass-approvals-and-sandbox, so Codex also runs outside its sandbox. Bypass Antigravity permission prompts passes --dangerously-skip-permissions to an Antigravity session. The tool set still bounds Daintree actions. Daintree confirmations separately chooses Follow agent setting (default, which follows the Assistant agent's own skip-permissions setting, or the global switch where that agent inherits it), Always ask, Never ask, or Ask, except for… a short list of actions you tick. Assistant sessions are not agent terminals.

See Daintree Assistant for what the assistant can do, how runbooks work and its permission settings, and MCP Server for the server itself.

Integrations group

CLI agents

A General subtab plus one subtab per built-in agent. Daintree ships 17 built-in agents. See AI Agents for the roster and each agent's specifics.

General holds the default agent (used by the help dock button and automated workflows like "What's Next?" and onboarding), a link to re-run the setup wizard, and the two global switches below.

Each per-agent subtab holds a Pin to toolbar toggle, expandable CLI help output, installation guidance with copyable commands and a re-check button when the CLI isn't detected, an agent-specific keyboard shortcut row, and the runtime settings card described under Presets and scopes.

Skip permissions

Permission bypass is layered, not a single checkbox, because "trust this agent everywhere" and "trust this one preset" are different decisions.

At the top sits a global switch, Skip permission prompts, under All agents on the General subtab. It is off by default, and first-run setup asks about it explicitly: the wizard's Agent permissions step presents the same toggle, defaulted off, and only writes it when you press Continue. Silence is not consent. The switch applies only to agents whose CLI actually supports a bypass flag, and it is read live at launch: it never rewrites per-agent state, so turning it off stops injecting bypass flags on the next spawn, restart, or resume.

Below it, each agent and each preset carries a three-way choice rather than a boolean:

ValueMeaning
DefaultDefer to the level above: preset to agent, agent to the global switch. The control shows what that currently resolves to.
OnForce bypass on for this scope.
OffForce it off. An explicit Off vetoes an On from any broader level.

The veto direction is the point: an explicit preset value wins over the agent value, and an explicit agent value wins over the global default, so a single agent or preset can be pinned safe without unwinding the global setting. When bypass ends up effective for a scope, the editor shows the exact flag it will add to the launch command: --dangerously-skip-permissions, --yolo, --dangerously-bypass-approvals-and-sandbox, and so on, depending on the agent.

Warning
Skip permissions passes the agent's own bypass flag. Its effect depends on the CLI; Daintree does not sandbox or intercept the agent's filesystem and network work. It removes the agent's own permission prompts, which are the point where a risky command waits for you before it runs. Turn it on for scopes you have deliberately decided to trust, not as a default. See Security & Privacy for what an agent can reach, and AI Agents for per-agent behavior.

Alt-screen mode

Agent CLIs can render either inline (output lands in Daintree's scrollback like any other terminal output) or on the terminal's full-screen alternate buffer, which is what the CLI does natively. Inline is smoother: WebGL scrollback, clean resizing, searchable history. Alt screen matches what you'd see running the CLI in your own terminal.

A global switch, Use alt-screen mode, sets the baseline (off, meaning inline). Each agent and preset then carries the same three-way Default / Inline / Alt screen control, resolving per agent through preset, then agent, then a curated per-agent registry default, then the global switch. As with skip permissions, an explicit choice at a narrower scope wins.

Presets and scopes

Each agent's runtime configuration is layered. The top of the Runtime settings card is a preset selector. The default scope, Default (all worktrees), applies everywhere. Everything else in the picker overrides it: project-shared presets loaded from .daintree/presets/ in the active project, CCR routes if you have Claude Code Router installed, and custom presets you create with Add preset.

Below the selector, a scope banner names what you're editing. For the Default scope and any custom preset the editor exposes custom arguments, skip permissions, alt-screen mode (on agents that support it), a fallback chain for when a provider is unreachable, and an env block. The Default scope's env block is the agent's global vars; a custom preset's is Env overrides, with the Default scope's vars shown inline as inherited rows.

In a custom preset, every field you've overridden shows a small reset icon. Clicking it drops the override and the field returns to inheriting. Project-shared and CCR presets are read-only, but Duplicate forks one into an editable custom preset.

Choose default settings or a preset from its source group.

Code Forge

One tab for every forge, replacing the separate GitHub and Forge tabs. A provider dropdown at the top switches between a General subtab and one subtab per registered provider. Provider subtabs are generic: GitHub is one provider among the registered set rather than a special case, and a provider without a settings tab of its own gets a credential form built from the fields its plugin declared. Saving validates the credentials against the provider before storing them, except for Beads, which can only check the token's format because each project names its own server. The GitHub subtab also holds a GitHub Enterprise Server section with its own host and token, and the Jira subtab chooses between Jira Cloud and Server or Data Center. Linear and Beads use the generic form, for a personal API key and a server token; their per-project settings live in Project settings > Plugins.

General holds three things:

  • Default forge provider: either "No global default (auto-detect from hostname)" or a specific provider to pin
  • Active project routing: one card per git remote of the active project, each with a badge showing which path resolved: Override, Default, Hostname, or No match, and a tooltip explaining why
  • Forge audit log: every provider call (list, get, assign, validate) with redacted argument summaries, for triaging slow providers and failure clusters
Note
Resolution order per remote is: per-project override, then global default, then hostname match, then no match. The badge tells you which won. Set the override in project Code Forge settings.

If the provider list fails to load, the tab raises the amber validation badge until it clears. Deep-linking to a provider whose plugin has since been disabled falls back to General rather than rendering a panel for a provider that isn't there.

See Code Forge for what a configured provider unlocks.

Hostname matching recognizes both GitLab and GitHub; upstream is in use.

Integrations

Two sections: External editor and Image viewer. Both are saved per-project, so the panel needs a project open before you can configure it.

External editor

Daintree opens files at a specific line and column in your editor of choice. Eight editors are built in (VS Code, VS Code Insiders, Cursor, Windsurf, Zed, Neovim, WebStorm/IntelliJ, and Sublime Text), plus a Custom option that takes a command and an arguments template using {file}, {line}, and {col} placeholders (default {file}:{line}:{col}). Quote the file placeholder if your paths contain spaces.

Discovery runs when the panel opens, scanning PATH and editor-specific directories, and lists each editor's status and resolved path. On macOS it also probes /Applications and ~/Applications for .app bundles, because GUI apps there inherit a minimal system PATH rather than your shell configuration. An editor that works fine from your terminal can otherwise show as (not found). Re-scan re-runs discovery without restarting Daintree, which is what you want after installing an editor mid-session.

Save persists the choice to the current project. Test opens your home directory in the configured editor and reports success or failure inline; it confirms the editor launches, not that your arguments template passes line and column correctly.

Three surfaces use it: Command Control Control +click on a file path in terminal output (a plain click opens the in-app viewer instead), the Open in Editor button in the File Viewer header, and the same button from a diff opened out of the Review Hub. If the configured editor is unavailable, Daintree falls back through $VISUAL/$EDITOR, then any detected known editor, then the platform open command, then a last-resort shell open. Line and column targeting survives only when a supported editor handles the request.

Image viewer

Which application opens image files. Defaults to the OS viewer; select Custom to supply your own command with {file} as the path placeholder. Per-project, like the editor.

Voice Input

Two sections: Speech-to-text and AI text correction.

Speech-to-text starts with a Transcription provider picker, the backend that turns speech into text:

  • OpenAI (default): Live Transcribe, with keyword biasing and spoken-command formatting
  • Deepgram: Nova-3, with server-side turn detection

The picker changes which API key field is shown, and a note beneath it spells out the data flow for whichever provider is selected: audio streams over an encrypted connection to that provider using your key. The settings include provider data-flow guidance; retention and training terms must be checked against the selected provider and account. The rest of the section:

  • Microphone permission status with platform-specific instructions, and a microphone selector
  • Advanced (collapsed) exposes an OpenAI organization ID and project ID, needed only for legacy user keys. The fields stay disabled for project-scoped sk-proj- keys
  • Language, and Paragraph breaks (spoken commands or manual Enter; spoken commands require English and fall back automatically)
  • Recording mode: Toggle (press the shortcut to start, press again to stop) or Push to talk (hold to record; releasing stops without submitting)
  • Custom dictionary of up to 100 domain terms sent to the transcription service to bias recognition, plus Learn words from corrections: when you fix a mishearing before sending, the term is offered as a dictionary suggestion you can accept or dismiss

AI text correction is optional post-processing that fixes technical terms, punctuation, and filler words. The model is fixed rather than chosen: there is no correction-model dropdown. It uses your OpenAI key even when transcription is running on Deepgram. It offers Resolve file references (turning "link to the input component" into an @file reference), Custom instructions appended to the core prompt, and an expandable view of that core prompt so you can see what your instructions are being appended to.

See Voice Input for the dictation workflow.

Hosts

Remote hosts lists the machines this one opens projects on over SSH, with Add host and a details page per host for renaming it, changing its SSH target, updating its build, its notifications and plugins, or removing it. Cloud providers holds the provider accounts Daintree creates cloud machines on. On macOS and Linux, This machine as a host, whose Allow this machine to be a host switch lets your other machines use this one, completes the tab. On Windows, the list holds WSL distributions on the same PC, and cloud machines and Host mode aren't available. The tab belongs to the machine you are sitting at, so a window on a host still shows this machine's host list.

See Add a Remote Host and This Machine as a Host.

MCP server

A local server that lets external MCP clients invoke Daintree actions. It binds only to 127.0.0.1 on the machine running the server: this machine for local work, or the host for a window on a remote host. In a window on a remote host, this tab shows and changes the host's server, the one the host's agents use.

Enable MCP server starts and stops it. The status and endpoint appear directly underneath, followed by expandable External clients and Internal connections lists. An external client can be disconnected from its row.

Connection holds the client-specific config, Port and API key, with Rotate API key below. Automatic port selection starts at 45454 and tries higher ports if needed. Copy a Claude Code JSON config, Codex TOML config, or plain endpoint values; Copy config for this project binds a client to the current workspace.

Audit log holds Capture audit log, Records kept and an Advanced diagnostics disclosure. Turn outcomes is a separate diagnostics section. The Assistant tab has its own audit privacy controls. See MCP server and MCP authorization for setup, tiers, limits and confirmations.

Plugins

Global installation management opens in the plugin manager (install from a file or URL, enable, disable, uninstall, check for updates), and this tab is a launcher into it, showing the installed count and an Open plugin manager button.

This global tab also offers Check for plugin updates in the background, off by default. Turning it on checks URL-installed plugins roughly once a day and adds an inbox notification when updates are available.

Project-specific controls live in Project > Plugins: trust and enable repository plugins, or control where installed plugin surfaces appear. See Plugins for installation and authoring.

Plugin actions

Plugins dispatch Daintree actions, and this tab is where you read what they dispatched. It holds a Record plugin actions toggle, a record cap, and an audit log viewer with copy, export, and clear. This records actions dispatched through Daintree's plugin bridge, not arbitrary filesystem or network work performed by a plugin worker or renderer view. Declared capabilities describe intent and do not sandbox plugin code.

Canopy

The Reading section has two switches. Show Canopy controls its toolbar button, shortcut and command palette entry. Hiding it also stops reading. Read agent terminals turns terminal readings on, and is on by default; when off, Canopy stops reading and deletes its readings from this computer.

Reading happens on this computer, with a small model that ships inside Daintree, and nothing it reads leaves it. Model shows the revision in use and the progress of an update, which Canopy checks Hugging Face for once a launch while reading is on. See Canopy's data flow.

Decisions

One switch, Enable Decisions, then the sections that appear once it's on: Connection (the relay, the recovery kit warning and the setup check), Devices and Remote Hosts, Encryption once a relay address is saved, Agents (a switch per supported agent, all off by default, and hidden in a window on a Remote Host), Questions, Policy, Decision model, Voice and Data. It's one setting for this computer, with no per-project layer. See Daintree Decisions.

Questions holds Default expiry, Show agent status and recent terminal lines on my phone (off), Hold phone notifications while I'm at this computer (on), Inbox with Open inbox, which opens the Decisions inbox, and Answers needing attention with Review.

Run history

A durable log of recipe runs and fleet broadcasts. Each entry shows what ran, against which worktree, and how it went: count pills for terminals spawned, failed, and defined, with per-terminal error text on the failures. Clear history wipes it behind a confirmation.

See Recipes and Fleet for what produces these entries.

Support group

Troubleshooting

Four sections:

  • System: Hardware acceleration (turn off GPU acceleration if you see blank panels or rendering artifacts) and System health check, which verifies Git, Node.js and npm are reachable
  • Diagnostics: Diagnostics report (Download diagnostics collects a snapshot of your environment, app state and recent logs, and you review it before anything is saved) and CPU profile, which captures 15 seconds of the interface's CPU activity to a .cpuprofile file for diagnosing lag
  • Logging: Application logs (open the log file, or Clear logs), Verbose logging for this session, Verbose logging on every launch, and Per-module log levels with its active overrides
  • Developer tools: Developer mode, which turns on two dependent switches, Auto-open diagnostics dock at startup and Focus events tab when it opens
Note
The plain Verbose logging switch is session-only and resets on the next app restart; set DAINTREE_DEBUG=1 in Daintree's launch environment to keep it.

See Troubleshooting and Diagnostics.

When a setting fails to save

Settings writes are optimistic: the control moves as soon as you click it, and the write to the main process happens behind it. When that write fails, Daintree rolls the control back to its previous value and says so where you are looking: most tabs show a Couldn't save that change banner in the section, noting that the setting is back to its previous value, with a Retry. A toggle that silently snaps back is a bug report; a toggle that snaps back and tells you why is a state you can act on.

The same principle covers reads. A tab that can't load its stored values shows an inline error banner with a Retry button and holds its controls rather than presenting an empty form you could save over real data. Ten tabs report their error state to the sidebar, so the validation mark appears on the tab even when you're looking somewhere else: Appearance, Worktree, Environment, Code forge, Voice input, Links, MCP server, Daintree Assistant, and the project Variables and Worktree setup tabs.

Note
On macOS and Linux, Daintree's persisted settings and runtime data are written owner-only (mode 0600, or 0700 for directories), and files created before that behavior landed are tightened at startup. These files hold API keys and environment variables in plaintext, so file permissions are the boundary. See Security & Privacy.