v0.9.19

Just resolving the conundrum with the previous release. Enjoy!

v0.9.18

Highlights

• Export and import projects. Project ▸ Export… saves the open project as a single .zip with its knowledge base and every task that isn't closed, plus the sprints, tags and statuses those tasks use. Project ▸ Import… turns that archive into a new project: you name it and pick its folder and agents. Archives are checked before anything is written, existing files in the folder are kept, and a failed import cleans up after itself so you can try again.
• Change a running task's permission mode live. The terminal header now has a permission-mode selector. You can move a run from Plan to Auto or Bypass, or between Auto and Bypass, without restarting it, on Claude, Codex and Pi (Pi 0.84.1 or later). The header shows "Switching to…" until the agent confirms the change and tells you plainly when a switch can't be applied. Claude's Auto mode now uses Claude Code's own auto mode when it's available, and falls back to accepting project edits when it isn't.
• Cowboy coding chains. Task Edit has a new Cowboy-coding checkbox, shown while the task runs in Bypass. When such a task finishes, the agent starts the next tasks automatically: every task that depended on it and is now unblocked, or otherwise the next task in the same sprint.
• "Thank You" voice mode. Settings ▸ Voice has a new Mode option. In Thank You mode you dictate freely and finish by saying "thank you" (or «дякую» in Ukrainian). The whole dictation is then cleaned of filler words, false starts and pause artifacts and inserted in one go. A "thank you" in the middle of a sentence doesn't end the recording. Standard mode works as before.
• New recording widget. While you record, a card at the bottom center shows a timer, a live waveform that follows your voice, and a Stop button that ends every active recording.

Improvements

• The board's bulk action bar now stages your changes. Move to, Status, Agent and Assignee are saved only when you click Apply. Reset, or closing the bar, discards them. When the staged agent is Claude, a Design mode toggle appears next to the agent picker.
• On macOS, the first launch explains and then asks once for notification, microphone and folder access (Desktop, Documents, Downloads and your open projects), so an agent never stalls mid-task on a system prompt. You can change these later in System Settings.
• Task and project chat replies have Copy buttons: one on each code block and one on every message.
• New chats appear in the left menu as soon as they start, without reopening the task.
• The Settings dialog has a cleaner, consistent layout across all tabs, with titled sections, aligned labels and controls, and switches for checkboxes. It keeps the same size when you switch tabs, so the tab strip no longer jumps around.
• The New Task Preview now shows the same card as the board. The title wraps properly, nothing gets clipped, and the sprint label reads correctly.
• Avatar, effort and agent badges on board cards and chat preview cards no longer cover tags, descriptions or status pills.
• Status names, Settings ▸ Workflow status names and dropdown arrows are now readable in every theme, light and dark.
• Dropdown arrows stay visible in Windows high-contrast themes.
• Finished or canceled tasks no longer raise idle notifications.

Fixes

• Fixed task and project chats that could sit on "Agent …" forever. Every turn now ends or shows its error, Stop frees the chat immediately, and Start Chat no longer reopens a broken chat.
• Codex chats now recover after a model-capacity or API error, so you can send another message instead of being stuck.
• Replying to a Codex plan with feedback no longer counts as approval. Later revised plans are announced and can be approved as expected.
• Saving from the New Task Preview or accepting chat proposals no longer overwrites changes someone else made to those tasks in the meantime.
• Saving a new task with no agent selected no longer fails.
• Clicking the agent badge on the pinned Discuss card in an objective chat now opens the task read-only instead of the agent assignment dialog.

---

Installers for macOS (Intel and Apple Silicon dmg), Windows (exe) and Linux (AppImage, deb and pacman) are attached below, with a SHA256SUMS file covering every installer.

v0.9.17

Highlights

• Agent permission modes. Choose how much freedom a coding agent gets on each task: Plan (the agent plans first and implements after you approve), Auto (the agent asks only when unsure) or Bypass (no approval prompts; the task runs unattended). Set a project-wide default under Project Settings → Workflow → Agent permissions, and override it per task in the Task Edit dialog.
• Image-generation tasks with OpenRouter. Add an OpenRouter image model in the model form and assign it to a task. The task description becomes the prompt, and the images are saved to a folder in your project.
• Objectives keep their source material. Tasks planned from an objective now keep its links, images, copy text, example values and data hints, so nothing from your original request is lost along the way.
• Banner clicks bring the app forward on macOS. Clicking a card on the always-on-top banner now shows the project window even when it is on another desktop (Space) or behind a full-screen app.

Improvements

• Each task card shows a shield icon for its effective permission mode. Hover it to see the mode, or click it to open Task Edit with the permission field focused.
• Running sessions in the side rail show the same shield next to the agent icon, including when the rail is collapsed.
• The terminal header shows which permission mode a run uses and whether it came from the project default. The mode is fixed when a run starts, and changes take effect on the next run.
• If an agent can't run in the chosen permission mode, the task doesn't start and the terminal explains why.
• Plan mode works the way it did before. The agent presents a plan, waits for your approval, then carries on in the same session. You can approve from the Task App or directly in the agent's own terminal dialog.
• Claude Code uses its built-in permission prompts again, and tasks start faster as a result. Bypass runs no longer stop at Claude's extra confirmation screen.
• Design Mode now works with any permission mode.
• In Bypass mode, agents make sensible decisions on their own instead of stopping to ask routine preference questions. They still ask when information is genuinely missing or when you request a consultation.
• Image model settings:
  • The model form has a Text or Image purpose choice, OpenRouter image model suggestions, aspect ratio, image count and an output folder.
  • Each image task sends a single generation request, with no automatic paid retries.
  • Image models are left out of the text-chat and planning model lists.
• When an objective becomes a single task, that task is treated as the full specification. Every link, image and detail from the objective counts as a requirement.
• If a generated task leaves out a design link or image reference from your objective, it is added back to the task description when you save.
• Project preferences now come from your signed-in account. The project creator sets the shared defaults, and other members keep their own personal settings. A local copy is kept per account and server, and it never contains credentials.
• All text on completed task cards is now an evenly muted grey, so finished work is easier to tell apart while staying readable in every theme.

Fixes

• The New Objective conversation now scrolls to the latest message whenever a new one arrives, instead of jumping back to the top.
• The permission mode menu is no longer cut off in a short Task Edit dialog.
• Fixed tasks sometimes missing from the board, and running sessions showing an outdated completion status in the side rail.
• Fixed the always-on-top notification banner showing no cards.
• In Plan mode, you can again answer the agent's plan and permission prompts from the keyboard in the terminal.
• Image task buttons now have a clearly visible keyboard focus outline in dark themes.
• Settings → General → Notifications no longer shows an empty yellow hint box with a blank button.

---

Installers for macOS (Intel and Apple Silicon dmg), Windows (exe) and Linux (AppImage, deb and pacman) are attached below, with a SHA256SUMS file covering every installer.

v0.9.16

Highlights

• Windows upgrades work again. Installing over an earlier version no longer fails with "TipATask cannot be closed" when nothing is running. Setup also replaces older installations cleanly and closes a running app more reliably.
• Pi tasks start on Windows. The startup check no longer times out on a cold start or on slower machines. If the bundled runtime can't be used, it falls back to an installed Node.js. When the check does fail, the message says which step failed and how long it took.
• Codex tasks on Windows receive their kickoff reliably. The app now waits until Codex has finished starting before it submits the task prompt. If the prompt gets stuck in the input box, the app retries it.
• A steadier board. Scroll position and the open card are kept across background refreshes, and the board no longer shifts while you drag a card.
• Safer quitting on Windows and Linux. Closing the last window now asks before quitting. Cancel keeps the window and its sessions open.

Improvements

• Windows installers are now available for both x64 and ARM64. Each includes the matching task terminal runtime.
• Claude's folder-trust prompt is pre-accepted for your projects, so a project's first Claude task goes straight to work.
• The planner now explains when it proposes no task changes, or when its proposed cards were rejected (for example, because the target tasks are locked). Before, these cases were treated as failures or retried automatically.
• Clearer messages when the planner returns an empty or unreadable response.
• When you answer a question dialog in a task or project chat, the answer now appears on the dialog only, not repeated as a separate chat bubble.
• When Pi is unavailable, the agent picker shows what was tried and what the check reported.

Fixes

• Starting a task from the edit dialog with a saved Pi model no longer reopens the agent picker.
• On the Backlog tab and on the board with Sprints turned off, a card no longer disappears after a status change until the next refresh.
• In the desktop app, active board filters and the Backlog count now update right after you confirm a status change.
• An open card stays in place when its task is updated or moves to another sprint, without leaving gaps or leftover overlays behind.
• Hover-opened cards no longer briefly collapse and reopen during refreshes. Cards with running sessions no longer re-animate their controls on every refresh.
• After a failed Claude chat turn, the question dialog reopens so you can answer again, and any text you typed is returned to the input box.
• Codex sign-in now works on Windows systems that don't allow symbolic links. Credentials are copied and kept up to date.
• Codex usage limits are now read correctly on Windows.
• Pi task prompts on Windows keep their quotes and line breaks, and long prompts no longer exceed command-length limits.
• Cancelling the quit dialog on Windows no longer leaves a hidden app running in the background that blocked the next launch.

---

Installers for macOS (Intel and Apple Silicon dmg), Windows (exe) and Linux (AppImage, deb and pacman) are attached below, with a SHA256SUMS file covering every installer.

v0.9.15

Highlights

• Ukrainian throughout the agent screens. The objective chat, Debug Console, connection banner, task terminal, agent selector, and messages sent by the server now follow the project language. This includes start failures, watchdog notices, and image-paste errors. English text is unchanged.
• Finish button for tasks an agent left unfinished. When an agent's terminal ends while its task is still in progress and the required commit is missing, the task is marked urgent and gets a comment. Its card also shows a Finish button, which relaunches the agent to commit and complete the task. The flag survives reloads, other windows, and app restarts. Idle agents are also reminded to commit and complete before they stop.
• Session limit that grows with your machine. A new default "ramp" admission mode raises the number of agent sessions one slot at a time while memory stays quiet. It drops back immediately when memory pressure appears. Queued tasks show how many sessions are running, the current limit, and when it may rise or why it is being held.
• Claude usage stays visible. The sidebar usage rows no longer go blank when the usage service is briefly unavailable or rate-limited. They show the last good reading, faded, with its time. That reading is kept across restarts. The app also checks the usage service less often to avoid being rate-limited.
• Better support for Pi models without tool calls. Custom endpoint models can be marked as not supporting tool calls, and such models show a "No tools" badge. Task and chat pickers hide these models. Objective planning still works with them, and turns that hit a tool-compatibility error retry without tools and show a clear message instead of a raw error.

Improvements

• Contrast across all themes now meets accessibility guidelines for accent text, focus rings, inline-edit outlines, tag chips, dependency lines, and selection rings.
• Completed and canceled cards and done To-Do rows stay readable. They no longer fade; a subtler background, dimmer text, and strike-through for canceled items mark them instead.
• The Start button on closed cards is now a clearly visible outlined button.
• The agent notice toast and several setup and chat elements now match light themes instead of showing a dark block.
• Saving a task from the New Task form or edit modal with a tag that does not exist yet now registers that tag automatically, with a short generated description.
• The New Task form keeps the sprint you picked, including Backlog and older sprints. Only a dependency may move it later, and the preview tells you when that happens.
• The New Task preview shows your project's task prefix and the key the task will actually get. It also checks dependencies on tasks outside the current board view.
• Task keys are now always assigned by the server, so tasks created by agents no longer get out-of-sequence numbers.
• Edit Agents shows the Claude Code and Codex model choices as cards with an inline editor, like Pi models.
• Re-checking an agent after unchecking all of them makes it the default again.
• Saved Pi model choices that can no longer run tasks now show a warning pointing to a model that can.
• The macOS notification banner no longer brings the Task App forward when you click Hide. Clicking a card does bring it forward.
• Merge conflict "Resolve with agent" now starts a fresh agent run when the previous session has already exited.
• Plans approved inside the agent's own terminal are now recognized, so the completion reminders and exit checks apply to those runs too.
• When an objective turn fails in the agent's command-line tool, the chat now shows a clear, localized error with next steps instead of a broken reply.
• Claude Code hooks now run in installed builds of the app. This includes knowledge-base auto-push, search result caching, and file-access tracking.
• Codex notification settings from a project's Codex config are carried into Task App Codex sessions without being lost.
• Agents are told to record follow-up tasks only for real defects, data risks, security issues, or steps you must take yourself. They comment on existing tasks before creating new ones.
• Lower background load: task changes are checked more cheaply, large server replies can be compressed, and idle project connections are released to save memory.
• Edits made while offline now appear in the window right away, and they replay correctly once the connection returns, including new tags. You are notified if the server rejects a queued change.
• When a reverse proxy reports the server as down, the app now treats this as offline and queues your changes instead of failing.

Fixes

• When Play All or bulk Start is refused by the server, you now see one message listing the refused tasks. Those tasks return to their previous status and agent, including starts that were refused after waiting in the queue.
• A single task start that the server refuses now returns the task to its previous status.
• Error messages are more accurate. "Network error" now appears only when the server could not be reached. Server rejections, such as a failed save or a comment you are not allowed to edit, now show their own reason.
• Opening a task you cannot access no longer starts a session or opens an empty objective tab.
• Watchdog memory figures now appear in whole MiB instead of long decimals.
• The Trello import error message now follows the current theme.
• The bulk action bar no longer hides behind expanded cards.
• The setup Agent step no longer lets you continue with no default agent. It also no longer quietly saves Claude as the default when you never picked it.
• Reading Claude usage is more reliable: there is more time for a slow start, and saved credentials are found more often on macOS.
• Reconnecting after the app starts while the server is down now retries properly.

---

Installers for macOS (Intel and Apple Silicon dmg), Windows (exe) and Linux (AppImage, deb and pacman) are attached below, with a SHA256SUMS file covering every installer.

v0.9.13

TipΔTask v0.9.13

v0.9.11

Windows compatibility problems

v0.9.10

TipΔTask v0.9.10

v0.9.9

Public Beta