Muster Agent 0.3.16

Chats on your own router keep their sub-agents and tools when you use a connector, Codex and computer use no longer need another app or npm on the Mac, your Claude Code skills and commands show up in Muster, and long chats compact at the right time.

• Code stays coloured everywhere: files up to the 512 KB read limit now keep their syntax colours in the file view, the edited-file and blame views, and while searching with Find (matches are marked on top of the colours, with the current one highlighted). Large files colour only the rows on screen, a block at a time off the main thread, so they open as fast as before. .cjs files and more file types (such as .json5, .ini, .less, .zig, .hs, .astro) are coloured too (#778).
• Every edit's inline diff opens as the whole file again, with its changes in place in the editor's Git colours: saved shell-edit chats show their diffs and changed-file totals again, live turns keep their streamed patches, and earlier edits of a file the chat edited again, edits made outside the chat's folder (for example in a worktree) and files the chat created no longer fall back to a few lines around the change (#781).
• Router models (planner, advisor and the other router agents) edit files with the native patch tool again, so their changes show as inline diffs and in the files-changed pill (#777).
• A patch a model runs through the shell shows as an inline diff and counts in the files-changed pill, instead of appearing only as a command (#779).
• A reply whose stream stops partway is saved with its full text, and a streamed piece without an id no longer hides later replies (#762).
• When a router refuses to continue a conversation, the chat carries on in a copy of it with the full history, instead of restarting from a short summary that lost the context (#767).
• Unanswered permissions now decline, Stop ends tool process trees, and startup leaves reused process IDs alone (#407).
• Chat titles keep their punctuation, New chat returns to your last project folder, matching chat names get a small suffix, and one-turn chats no longer show a stray minimap dash (#403).
• A temporary provider failure after work has started now gets up to two automatic continuations on the same thread, with a countdown and a notice for each failure. A turn that ends after a tool call without replying gets one chance to finish, and live plan updates show their checklist (#386).
• UI fixes from the 0.3.11 sweep (#398). Only one sidebar item is highlighted at a time, and Memory is highlighted while it is open. Two folders with the same name now show their parent folder in the sidebar and in search. Memory suggestions start as one line you can expand to edit. A turn's time no longer counts an import notice added later, so it no longer reads hundreds of hours. A gateway that hasn't been checked says Not checked yet instead of Ready. An empty project no longer shows In progress. The sidebar folds away below 960px.
• Workspace writes and edits can no longer leave empty files or folders outside the chat’s folder when another process swaps a parent folder for a link (#710).
• Chats on router models keep their sub-agents, file edits and MCP tools after you ask for a connector such as Figma or Google Drive. Muster now loads only the connectors the chat named instead of every connector's tools, which the router cut off past 128 (#756).
• MCP credentials stay out of process command lines, saved API keys are cleared when their endpoint changes host, and release builds require a signing identity and pinned build actions (#406).
• Chats on API-key and local router routes remember earlier tool results, fit requests to the model's window and retry brief outages without repeating commands. Long command output keeps its opening and final summary, files can be searched and read in pages, and the agent only gets instructions for tools it can use (#384).
• In a narrow window the browser preview card now sits above the chat instead of covering its text (#637).
• Installing Codex from Muster no longer needs npm, Node or another app on the machine. Muster downloads Codex's own build for your Mac, checks it against the published checksum and runs it directly, so a clean machine can sign in and chat (#721).
• Your Claude Code skills and commands now show up in Muster. Skills from ~/.claude/skills and a project's .claude/skills, and commands from ~/.claude/commands and .claude/commands, appear in the / picker and the + menu. Picking a command inserts /name and runs it, and the chat shows the prompt it ran. Plugins and MCP servers also show their own logos across Skills & plugins, Settings › MCP servers, the + menu's Connectors & MCP section (now in New chat too), composer chips and tool rows. Plugins that ship without a logo pick one up once and keep it cached, and a logo that can't be drawn falls back to the plugin's initials. When a project command shares a name with one of yours, or a command with a skill, the one you picked is the one that runs; typing /name by hand runs the same one Claude Code would. A quoted argument such as "Button primary" stays one argument, as in Claude Code.
• Long chats now compact before they fill the window the model really runs with, and the context meter shows the right percentage. A chat on a router model listed at 1M but run at about 258K used to read 15% when it was 58% full and overshoot before compacting. Switching a chat to another model no longer carries the old model's window over (#739).
• A Codex route set up in config.toml now starts on its own provider when Muster itself was opened from a Muster terminal; it used to pick up that terminal's profile and fail to load (#740).
• Claude models on your own router now run through Claude Code when it is installed, pointed at the router with the route's own key. They no longer go through the coding client, so the repeated "Reconnecting…" and "stream closed" failures are gone, and your own tool servers, approvals and steering work as they do with the Claude Code sign-in. GPT models, connector turns and goals stay on the coding client.
• Claude chats on your own router no longer reset to a summary when the router has lost the saved thinking for an earlier tool call. Muster resends that part of the history as plain text, so the chat keeps its full history and thread and carries on. Starting fresh is now a last resort.
• Chats on your coding-client sign-ins and profiles no longer show up in your other coding apps. Muster now keeps its own threads in its data folder while sharing your sign-in, profiles, skills and plugins; existing chats continue where they left off.
• Switching a chat to a model on another provider or account no longer loses the conversation. The first message on the new model now carries a summary of the earlier messages, as forks and compaction already do, and the chat says so.
• Computer use no longer needs another desktop app installed. Muster ships its own computer-use helper for macOS: every agent chat, on Codex, Claude Code or your own router, can read an app's controls, click, type, press keys and scroll. Each step answers with only what changed on screen, so runs are quicker and use less context; read-only chats can look but not act. An orange Muster cursor shows where the agent is working, in the browser and in desktop apps, and fades out when it stops.

Download

Mac	File
Apple silicon (M1 and later)	Muster-Agent-0.3.16-arm64.dmg (or .zip)
Intel	Muster-Agent-0.3.16-x64.dmg (or .zip)

Open the disk image and drag Muster Agent to Applications.

Windows	File
Windows 10/11 (x64)	Muster-Agent-0.3.16-win-x64-setup.exe (or the portable -win-x64.zip)

Run the installer; it installs for your user, no admin needed. It is not code-signed yet, so Windows SmartScreen may warn: choose More info → Run anyway.

Linux	File
AppImage (any distro, x64)	Muster-Agent-0.3.16-linux-x86_64.AppImage
Debian / Ubuntu	Muster-Agent-0.3.16-linux-amd64.deb
Archive	Muster-Agent-0.3.16-linux-x64.tar.gz

AppImage: chmod +x the file, then run it. Debian/Ubuntu: sudo apt install ./Muster-Agent-0.3.16-linux-amd64.deb.

Muster Server (self-hosted)	File
macOS, Apple silicon	muster-server-0.3.16-darwin-arm64.tar.xz
macOS, Intel	muster-server-0.3.16-darwin-x64.tar.xz
Linux x64	muster-server-0.3.16-linux-x64.tar.xz

Extract it, then bin/muster-server init and bin/muster-server start. It includes its own Node runtime. Docker, systemd and TLS setups are in docs/server.md.

Windows (installed with the setup.exe), AppImage and Debian/Ubuntu installs update themselves from Settings › Updates, downloading only what changed where they can.

Checksums for every file are in SHA256SUMS.

This build is not notarized by Apple (signing: self-signed). The first time, right-click Muster Agent in Applications and choose Open, or run:

xattr -dr com.apple.quarantine "/Applications/Muster Agent.app"

Muster Agent needs a model provider sign-in; it detects the ones already on your computer on first run.

Muster Agent 0.3.15

Claude chats on your own router recover on their own when an older conversation can no longer be continued, and notification toasts move to the bottom-right corner, clear of your messages.

• Claude chats on your own router that continue an older conversation no longer fail with "replay state expired or is unavailable". When the router has dropped the earlier conversation, Muster says so, starts a fresh one with a summary of the earlier messages and sends your message once more.
• Notification toasts now stack in the bottom-right corner of the window, just above the status row, instead of the top-right where they sat over your messages. They stay clear of the message box and the chat summary card; in a narrow window they lift above the message box, and with the right pane open they stay in the window corner and follow the status row when it moves.

Download

Mac	File
Apple silicon (M1 and later)	Muster-Agent-0.3.15-arm64.dmg (or .zip)
Intel	Muster-Agent-0.3.15-x64.dmg (or .zip)

Open the disk image and drag Muster Agent to Applications.

Windows	File
Windows 10/11 (x64)	Muster-Agent-0.3.15-win-x64-setup.exe (or the portable -win-x64.zip)

Run the installer; it installs for your user, no admin needed. It is not code-signed yet, so Windows SmartScreen may warn: choose More info → Run anyway.

Linux	File
AppImage (any distro, x64)	Muster-Agent-0.3.15-linux-x86_64.AppImage
Debian / Ubuntu	Muster-Agent-0.3.15-linux-amd64.deb
Archive	Muster-Agent-0.3.15-linux-x64.tar.gz

AppImage: chmod +x the file, then run it. Debian/Ubuntu: sudo apt install ./Muster-Agent-0.3.15-linux-amd64.deb.

Muster Server (self-hosted)	File
macOS, Apple silicon	muster-server-0.3.15-darwin-arm64.tar.xz
macOS, Intel	muster-server-0.3.15-darwin-x64.tar.xz
Linux x64	muster-server-0.3.15-linux-x64.tar.xz

Extract it, then bin/muster-server init and bin/muster-server start. It includes its own Node runtime. Docker, systemd and TLS setups are in docs/server.md.

Windows (installed with the setup.exe), AppImage and Debian/Ubuntu installs update themselves from Settings › Updates, downloading only what changed where they can.

Checksums for every file are in SHA256SUMS.

This build is not notarized by Apple (signing: self-signed). The first time, right-click Muster Agent in Applications and choose Open, or run:

xattr -dr com.apple.quarantine "/Applications/Muster Agent.app"

Muster Agent needs a model provider sign-in; it detects the ones already on your computer on first run.

Muster Agent 0.3.14

Chats keep working through quits, crashes and usage limits; Claude on your own router works in Full access; secrets are masked everywhere a chat is saved; and many fixes to queues, receipts, files, git and right-to-left text.

• A side chat now has a Back to chat button in its header that returns to the conversation it was opened from, restores a maximized resource pane and focuses the message box. The button is hidden when the side chat has no parent conversation.
• Large files stay quick to open when the last turn edited them or when you turn on Blame: only the lines near the view are drawn, and the change markers, Keep/Undo, Next/Previous change, copy and find keep working.
• Receipts no longer count files a checkout, merge or rebase brought in as the turn's own edits: when the branch moves during a turn, only files the agent changed or committed are listed. A tool the provider never closed is shown as Stopped once its turn ends, and after Muster restarts, instead of running forever.
• Secrets you saved in Muster, and environment variables named like credentials (*_TOKEN, *_API_KEY, *_SECRET, *_PASSWORD…), are masked as ••••(NAME) wherever a chat is saved: the thread, tool output logs, chat titles and exports, including chats exported after the secret was saved.
• After a crash, a message you had queued is no longer sent twice when you choose Continue anyway and then Steer: it goes once, and the saved copy Codex kept for it is deleted.
• Files written by an API-key or router chat in Workspace access can no longer be pushed outside the chat's folder by a folder or link swapped during the write, and a swapped folder no longer leaves empty folders outside.
• The OpenAI (ChatGPT sign-in) model list no longer hides newer models when a profile sets an older model_catalog_json: models your account gained since, from Codex's own model cache, are listed after the catalog's. Hidden models and other providers' ids stay out.
• Claude models on a router you configured (for example cc/claude-opus-5-5, or the router's planner and advisor agents when they run Claude) no longer fail with "thinking blocks cannot be modified" errors in Full access chats. These turns now talk to the router's Messages API with Muster's own shell, file and browser tools, and every thinking block goes back exactly as received. Workspace and Read-only chats, chats in a sandbox, and turns that use Project, mailbox, terminal, SSH or your own tool servers still run through the Codex client. The in-app browser works on this route; canvases, background commands and computer use are not offered on it. Moving a chat between the two ways of running continues from a summary of the earlier messages.
• A task with a very long thread (thousands of messages) now still wakes its agent: the agent gets the newest messages and a count of the older ones, and the task activity shows how big the packet was. A long prompt now goes to Gemini through standard input instead of the command line. If a task fails to start with the same error three times in a row, it stops retrying and one Problem appears in the Inbox until you resolve it.
• Sub-agents are now named after what they were asked to do ("Count the lines in data/lines.txt") instead of the first words of their prompt, so three agents started in the same folder no longer show three identical rows. Names are unique within a chat, and the Subagents panel counts finished agents as completed.
• If Muster fails part-way through starting, or wakes from sleep after shutting down, the automations scheduler no longer runs against a closed database.
• The agent's rule about not stopping processes it did not start now applies to local processes only. Remote hosts, systemd or launchd services, and anything you explicitly ask it to stop or restart are no longer refused.
• In Full access the agent no longer ends a turn asking whether to push, commit, install or continue; it just does it.
• Chats on API-key routes only ask before irreversibly deleting user data, not before push, install or restart.
• Switching a chat to another provider now always hands the new provider the task: the first request, your last five messages, the last assistant summary and the folder's git branch and status come first, before tool activity.
• After you stop a run, sending a new message no longer asks whether to clear the queue and never drops it: the held messages stay queued and run, in order, after the new one.
• If a run is stopped before your message was sent, your text comes back into the composer instead of disappearing.
• Memory suggestions no longer appear inside the chat transcript. They collect in the Memory inbox, and each chat lists at most one new suggestion per day (later ones are kept, not shown).
• Enter while a run is going now steers it by default (Cmd/Ctrl+Enter queues instead). Settings › Chat › "Enter while running" switches the default back to Queue.
• Steering is more reliable: on the Claude Code route a steer folds into the current task instead of ending it, a message an API route can no longer read is queued instead of being lost, and a Codex steer error is shown with the message queued.
• When a run stops on a provider usage or rate limit, the chat now shows Limited with the reset time (from the usage windows Muster reads, or the provider's own message) and continues by itself at the reset. Queued messages wait and then send in order. Each chat can choose "Resume at reset" or "Don't resume", and Settings › General › "Continue when a usage limit resets" turns it off everywhere. The wait survives a restart. If the provider still refuses, Muster backs off, says so once, and stops after a few tries instead of looping. A rate limit or outage (429, 503) is retried sooner than a usage window.
• A limited chat can opt in to a fallback model: pick one in the Limited row and the chat continues on it straight away, with the earlier conversation as a summary. Off by default.
• Quitting Muster no longer counts as stopping your runs. A turn that was working when Muster quit or crashed now continues after the next start: Codex threads and Claude Code (and other CLI) sessions resume the same session, other routes get a short "Muster restarted…" message. A chat waiting for your approval or answer is left for you, a chat that needs its provider status checked keeps "Continue anyway", and a continuation is never repeated if Muster closes again during it. Turn it off in Settings › General › "Continue interrupted chats after a restart".
• Installing an update no longer quits over running work: Muster says it will update when the work finishes, then installs.
• Launch and connection failures that never reached the provider are now retried five times with growing waits (1 to 16 seconds) instead of twice.
• Queued messages can no longer be lost or sent twice around a quit: a message leaves the queue only in the same step that records it as sent, a message already sent is dropped at launch, and messages the Codex app-server held come back to Muster's own queue.
• Tool rows no longer keep spinning after a Stop, a crash or a quit; they show as interrupted.
• Browser pages without their own background now render on white instead of the app's dark background.
• The browser preview reserves room beside the conversation instead of covering transcript text.
• Closing a terminal that already ended no longer fails, a terminal opened as its tab is dismissed is closed instead of left running, and terminal errors show plain text instead of the raw IPC message
• The agent can ask you a question without stopping: it shows the usual question card with the choice it is continuing with, and keeps working. Answer whenever you like; the answer reaches the running agent right away, or becomes the next message once the run has ended.
• Stop failed network reconnects after three attempts or 30 seconds, explain proxy, DNS, TLS and refused connections, and offer Retry.
• Security: the agent's background-process and terminal tools now prove which chat is calling with the same per-chat token as Muster's other agent tools, so one chat can no longer start, read or stop processes, or read a terminal, as another chat. A call whose chat and token don't match is refused with a plain message.
• The Ledger's chain check now also catches deleted or rewritten recent receipts, not only edits in the middle. It names the first receipt affected, and adding new receipts afterwards does not hide it.
• Installing a plugin or MCP connector while a chat is working now makes it usable right away: the chat restarts on a fresh session (keeping its conversation) and re-sends your last message, with a short notice saying so. Idle chats whose last message asked for that plugin get the same treatment; unrelated chats are left alone.
• Memory: facts and preferences you told Muster once now carry into project and folder chats, and recall looks at your task and recent messages (not just "continue"), with a longer wait for the memory service.
• Memory context is sent again on the next turn after a compaction or a model/provider change.
• The browser tab an agent is working in now stays usable when you close the browser panel mid-run: it is no longer frozen or discarded while in use, live previews revive it, and browser tool calls retry safely after a dropped connection (never repeating a click or keystroke).
• Project task runs now keep working while a build, test or deploy is running instead of stopping to wait, and only stop for decisions the user must make.
• Agents post the task handoff comment themselves instead of drafting it and asking whether to post it.
• A task you checked out and then left idle for days now shows up in the Inbox under Needs you, saying how long it has been idle, with Hand back, Release and Keep. The task on the server gets one note saying how long it has been idle. Nothing is handed back or released until you choose.
• Usage stats and Receipts now count...

Muster Agent 0.3.13

Replies in one turn read as one answer, and pictures that can no longer be read take one line instead of an empty frame.

• Consecutive assistant replies within one turn now sit at the paragraph gap instead of a wide gap.
• An image that cannot be read no longer leaves a large empty box in a reply's gallery: the hero collapses to one line, unreadable thumbnails are hidden, and a gallery where every image failed lists one line per image with no strip or Canvas button.

Download

Mac	File
Apple silicon (M1 and later)	Muster-Agent-0.3.13-arm64.dmg (or .zip)
Intel	Muster-Agent-0.3.13-x64.dmg (or .zip)

Open the disk image and drag Muster Agent to Applications.

Windows	File
Windows 10/11 (x64)	Muster-Agent-0.3.13-win-x64-setup.exe (or the portable -win-x64.zip)

Run the installer; it installs for your user, no admin needed. It is not code-signed yet, so Windows SmartScreen may warn: choose More info → Run anyway.

Linux	File
AppImage (any distro, x64)	Muster-Agent-0.3.13-linux-x86_64.AppImage
Debian / Ubuntu	Muster-Agent-0.3.13-linux-amd64.deb
Archive	Muster-Agent-0.3.13-linux-x64.tar.gz

AppImage: chmod +x the file, then run it. Debian/Ubuntu: sudo apt install ./Muster-Agent-0.3.13-linux-amd64.deb.

Muster Server (self-hosted)	File
macOS, Apple silicon	muster-server-0.3.13-darwin-arm64.tar.xz
macOS, Intel	muster-server-0.3.13-darwin-x64.tar.xz
Linux x64	muster-server-0.3.13-linux-x64.tar.xz

Extract it, then bin/muster-server init and bin/muster-server start. It includes its own Node runtime. Docker, systemd and TLS setups are in docs/server.md.

Windows (installed with the setup.exe), AppImage and Debian/Ubuntu installs update themselves from Settings › Updates, downloading only what changed where they can.

Checksums for every file are in SHA256SUMS.

This build is not notarized by Apple (signing: self-signed). The first time, right-click Muster Agent in Applications and choose Open, or run:

xattr -dr com.apple.quarantine "/Applications/Muster Agent.app"

Muster Agent needs a model provider sign-in; it detects the ones already on your computer on first run.

Muster Agent 0.3.12

The browser, steering, sub-agents and compaction work on every route, Paperclip tasks run locally with their full history, and approvals no longer expire behind your back.

• The in-app browser now works on more model routes. Claude Code, OpenCode and Grok chats load Muster's browser tools, and OpenAI-compatible HTTP routes (OmniRoute, OpenRouter, custom connections) get them as built-in tools that drive the same browser tab, limited to opening and reading pages in read-only chats. Each route is told the exact tool names. Gemini CLI, Cursor CLI and the chat-only Anthropic API route say plainly that the browser is not available there, instead of naming tools they do not have.
• A sub-agent can now pass its own session name to the browser tools and get its own tab, so agents working in parallel no longer fight over one page.
• Fewer "Muster's browser is not reachable … timeout" errors. The browser bridge starts again on first use if it failed at launch, a refused connection is retried once, and a page that hangs answers within 100 seconds with what to do next, instead of timing out.
• The right-sidebar browser now shows the page the agent is driving. Before, a chat could say "Clicked Validate" while the sidebar tab for that page sat on "Starting…" and "Disconnected": the agent's page had never been given a size, so no live picture was captured, and the sidebar opened a picture placeholder instead of the agent's own browser tab. Opening the agent's page from the picture-in-picture, the transcript or an older layout now opens its live browser tab, and the tab reconnects after a reload or a restart.
• Steering a running chat now works in Claude Code chats. Before, a message sent while Claude Code was working was passed on but the run then never finished, because Muster waited for an answer to every message while Claude Code folds them into the turn it is already doing. The run now ends normally, and you can steer several times in a row: each message reaches the agent in order and shows in the chat.
• Steering now also works on router and API routes (OpenAI-compatible, Anthropic and custom endpoints): your message joins the conversation before the model's next step, and later turns remember it.
• On routes that can't take a message mid-run (Gemini, Cursor, OpenCode, Grok), Steer and the Steer button on a queued message now say "Queued — this route can't take messages mid-run; it sends when this step ends" instead of doing nothing visible. The message stays queued.
• Messages steered in quick succession are sent one at a time, in the order you sent them.
• A Muster Server task you work on locally now starts with everything the server's agents already did, not just its title. Before the first turn Muster takes a fresh copy of the task and writes its whole history into a hidden .muster/task/ folder in the worktree or folder (never committed): the full description with its parent and blockers, the whole thread with authors and times, every document's current text, and the earlier agents' runs with their summaries, what they posted and the pull requests, branches and files they produced. The agent gets a short summary of what was done and what is still open, and is told to read those files first. Later turns pick up new comments and runs. If the server cannot be reached, the last copy is used and the agent is told so. Secrets are removed from all of it.
• When the server's agents worked on a branch that is on your repository's remote, the local worktree now starts from that branch, so their commits are already there. Otherwise the agent is told which branch or commit to continue from. Your own checkout is never switched or changed.
• The same applies to tasks from an imported project, whether you press Start or the scheduler starts them.
• Auto-dispatch on a project imported from Muster Server now runs each task in a worktree of the project's folder (on the task's own branch, reused next time), exactly as Start does, never in the folder you work in. If the worktree cannot be made, the run does not start and the project's activity says why.
• Sub-agents start again in normal chats, Project chats, check-out chats and task runs. Codex chats now launch with multi-agent turned on and use the chat's own model for sub-agents when your default sub-agent model isn't offered by that provider, instead of failing with "Executor model does not match the selected provider". Claude Code chats may always use the Task tool, whatever the permission mode. Routes with no sub-agent tool (API keys, routers, local servers, Gemini, Cursor and Grok CLIs) say so in one line when you ask for sub-agents.
• A project or task you deleted in Muster is no longer made again when you import from Muster Server. Muster remembers the deletion by the server's id, leaves the item out of every later import (the report says what it skipped), and lists it under "Deleted here" in the import list with a Restore button: Restore, then import again, and it comes back (a project with its tasks). Deleting never writes anything to the server.
• Thread decisions from Muster Server can now all be answered in place. Suggested tasks are listed and you create the ones you tick; a checklist approves the boxes you tick; item verdicts take one verdict per item (with a reason where the request needs one); a connection request names the app and the waiting agent and can be declined here. When a tool call needs approval, the card shows the tool, whether it reads, changes or destroys data, what it will do, its arguments, and afterwards whether it ran or failed. Failed, replaced and out-of-date requests are described plainly instead of "Answered." or a raw status, a request the agent could not resume says so, Approve is switched off with a reason while the server is still preparing the workspace, and a question marked optional can stay empty.
• Comments you send to Muster Server carry one id per draft, so a send that timed out and is sent again is posted once. Cost entries for local work carry a key too, so a post whose answer was lost is retried and counted once. Pictures attached on the server show in replies, fetched with your sign-in. Task lists are read by id cursor so no task is skipped when others change while loading, and a rate-limit answer is retried once after the wait the server names.
• A chat that was running when Muster quit or crashed is no longer stuck. Chats on routes that run inside Muster (Claude Code, OpenCode and other local routes) are simply marked interrupted, so you can send a new message straight away. When Muster cannot confirm what happened at the provider, the notice now offers "Continue anyway" next to "Check provider status", so you are never blocked.
• Approvals and questions now wait 2 hours instead of 10 minutes. A request that runs out of time shows as "Timed out" (not "Denied"), and "Ask again" tells the agent the request timed out so it can ask again. After a restart, requests that were open show "Lost when Muster restarted" with the same one-click "Ask again". Approving a request that has timed out, or whose step already finished, is refused instead of running something stale.
• Muster now sends a desktop notification when an approval or question is waiting for you, and one reminder about 5 minutes before it times out. Both follow your notification settings: nothing is sent while you are muted or have attention alerts off, and if you are muted past the deadline, the request stays open until your mute ends. Muster Server chat connectors (Slack and others) also post chat approvals and questions, not only hiring approvals. Chats in Full access still never ask for approval.
• The Paperclip Inbox now keeps agent questions and approvals on tasks you are accountable for, reads the whole attention feed instead of the first 50, shows critical items above high ones, finds @mentions that the 120-character activity snippet cut off, and clears a mention once you open the task. With a board key and no live connection, the Inbox badge re-checks once a minute (ETag revalidated, and not at all while the window is hidden or unfocused).
• Hand-back, release and "run on the server" now check the task on Muster Server first. If it was cancelled, closed or reassigned while you worked, Muster sends nothing, ends the check-out on this Mac, keeps your worktree and says so ("RAG-12 was reassigned to Bob on Muster Server; your local work is kept in ..."). This also applies to an automatic hand-back that was counting down when the app restarted. A refused sign-in (401) now shows "Sign in to Muster Server again" and waits for a new key instead of retrying as if offline. Unsent updates retry after 30 s, doubling up to 30 minutes, and are given up on after about a day, with "N updates couldn't be sent · Retry".
• Project decisions that ask a question now show Paperclip's "Other" answer: choose Other to type your own reply (the field is focused for you), or answer a text-only question directly. Once answered, the card shows "Other: your text".
• Automatic hand-back no longer fires on unfinished work. If the agent's last message ends with a question or an offer ("Should I commit item 1 now?", "Want me to...", "Would you like..."), nothing is committed or handed back until you answer or say you are done. Uncommitted work is now committed only after the 60 second countdown, not before it, and a running countdown is saved so an app restart picks it up again.
• Saying "I am done", "done", "done, hand back", "hand it back", "finished" or "all done" now hands the task back. When the hand-back is held because no tests ran, the task offers "Run tests and hand back" and "Hand back anyway".
• Checking the same task out again reuses its worktree and chat instead of making a "-2" copy. Any chat opened in the task's worktree counts as work on it, and "Paused - no activity" is no longer posted while any of those chats was active.
• A goal now closes only when the agent reports it com...

Muster Agent 0.3.9

The model picker now keeps up with new releases. Claude Haiku 5.5 is added (with Light to Extra High reasoning levels; Haiku 4.5 stays). GPT-6.1 Sol and any other model Codex lists as visible show without being switched on in Settings.

• The ChatGPT model list is read from Codex's models_cache.json with the same 16 MB limit as a model catalog. It used the 1 MB config-file limit, which a cache holding many models' full instructions can outgrow, hiding the whole list.
• When Codex rewrites its cache and one model replaces another (same count), the picker is now told. Before, Muster compared only the number of models, so the swap was never pushed to the open app.
• The Claude catalog (anthropic-models.json, dated 2026-10-12) is no longer masked by an old copy: an override file in the data folder or ~/.muster/ whose updated date is older than the one shipped is ignored. An override without a date is your own list and still wins. The live Anthropic /v1/models list is still added when an API key is set.

Download

Mac	File
Apple silicon (M1 and later)	Muster-Agent-0.3.9-arm64.dmg (or .zip)
Intel	Muster-Agent-0.3.9-x64.dmg (or .zip)

Open the disk image and drag Muster Agent to Applications.

Windows	File
Windows 10/11 (x64)	Muster-Agent-0.3.9-win-x64-setup.exe (or the portable -win-x64.zip)

Run the installer; it installs for your user, no admin needed. It is not code-signed yet, so Windows SmartScreen may warn: choose More info → Run anyway.

Linux	File
AppImage (any distro, x64)	Muster-Agent-0.3.9-linux-x86_64.AppImage
Debian / Ubuntu	Muster-Agent-0.3.9-linux-amd64.deb
Archive	Muster-Agent-0.3.9-linux-x64.tar.gz

AppImage: chmod +x the file, then run it. Debian/Ubuntu: sudo apt install ./Muster-Agent-0.3.9-linux-amd64.deb.

Muster Server (self-hosted)	File
macOS, Apple silicon	muster-server-0.3.9-darwin-arm64.tar.xz
macOS, Intel	muster-server-0.3.9-darwin-x64.tar.xz
Linux x64	muster-server-0.3.9-linux-x64.tar.xz

Extract it, then bin/muster-server init and bin/muster-server start. It includes its own Node runtime. Docker, systemd and TLS setups are in docs/server.md.

Windows (installed with the setup.exe), AppImage and Debian/Ubuntu installs update themselves from Settings › Updates, downloading only what changed where they can.

Checksums for every file are in SHA256SUMS.

This build is not notarized by Apple (signing: self-signed). The first time, right-click Muster Agent in Applications and choose Open, or run:

xattr -dr com.apple.quarantine "/Applications/Muster Agent.app"

Muster Agent needs a model provider sign-in; it detects the ones already on your computer on first run.

Muster Agent 0.3.11

Router and API-key chats work longer and stay quiet about sub-agents.

• Chats on routes that cannot start sub-agents (API keys, routers, local servers) no longer keep saying "I don't have a sub-agent tool in this chat". Sub-agent hints now appear only on routes that can use them, and those chats are told to just do the work.
• Router and API-key chats can work much longer in one turn (up to 400 tool rounds, 1,000 tool calls or 2 hours) before stopping. When a limit is reached, the model now gives a short status of what is done and what is next instead of a bare "Stopped" note.

Download

Mac	File
Apple silicon (M1 and later)	Muster-Agent-0.3.11-arm64.dmg (or .zip)
Intel	Muster-Agent-0.3.11-x64.dmg (or .zip)

Open the disk image and drag Muster Agent to Applications.

Windows	File
Windows 10/11 (x64)	Muster-Agent-0.3.11-win-x64-setup.exe (or the portable -win-x64.zip)

Run the installer; it installs for your user, no admin needed. It is not code-signed yet, so Windows SmartScreen may warn: choose More info → Run anyway.

Linux	File
AppImage (any distro, x64)	Muster-Agent-0.3.11-linux-x86_64.AppImage
Debian / Ubuntu	Muster-Agent-0.3.11-linux-amd64.deb
Archive	Muster-Agent-0.3.11-linux-x64.tar.gz

AppImage: chmod +x the file, then run it. Debian/Ubuntu: sudo apt install ./Muster-Agent-0.3.11-linux-amd64.deb.

Muster Server (self-hosted)	File
macOS, Apple silicon	muster-server-0.3.11-darwin-arm64.tar.xz
macOS, Intel	muster-server-0.3.11-darwin-x64.tar.xz
Linux x64	muster-server-0.3.11-linux-x64.tar.xz

Extract it, then bin/muster-server init and bin/muster-server start. It includes its own Node runtime. Docker, systemd and TLS setups are in docs/server.md.

Windows (installed with the setup.exe), AppImage and Debian/Ubuntu installs update themselves from Settings › Updates, downloading only what changed where they can.

Checksums for every file are in SHA256SUMS.

This build is not notarized by Apple (signing: self-signed). The first time, right-click Muster Agent in Applications and choose Open, or run:

xattr -dr com.apple.quarantine "/Applications/Muster Agent.app"

Muster Agent needs a model provider sign-in; it detects the ones already on your computer on first run.

Muster Agent 0.3.10

Previews and project chats behave. Browser and computer-use previews show only while a session is live and show the real page; Compact works in Project chats; router models show their context window.

• The browser and computer-use picture-in-picture (#360) now appears only for a live session. When a session ends it shows "Ended" with its last screenshot for a few seconds and goes away; a chat reopened later no longer keeps a ghost card, and unrelated shell work no longer leaves a red "disconnected" dot on an old page.
• A card without a picture shows a "Starting…" skeleton instead of a globe. The corner dot means live and has an accessible label. Clicking a card opens that session's existing tab, never a blank browser. The browser tab shows "Opening browser…" instead of an empty pane while it starts.
• Compact now works in Project chats. When no live provider session holds the thread (after a restart, a stopped turn, or a provider switch) Muster compacts from the saved transcript: the next message starts a fresh provider conversation seeded with a summary of the earlier messages, instead of failing with "No live provider session holds this thread".
• The context popover shows window size and remaining context for router models. The router's /models context_length / max_input_tokens is used when the provider reports no window.
• Typing /compact is always handled by Muster (every chat type and provider), never sent to the model. Compact errors no longer show the "Error invoking remote method" prefix.

Download

Mac	File
Apple silicon (M1 and later)	Muster-Agent-0.3.10-arm64.dmg (or .zip)
Intel	Muster-Agent-0.3.10-x64.dmg (or .zip)

Open the disk image and drag Muster Agent to Applications.

Windows	File
Windows 10/11 (x64)	Muster-Agent-0.3.10-win-x64-setup.exe (or the portable -win-x64.zip)

Run the installer; it installs for your user, no admin needed. It is not code-signed yet, so Windows SmartScreen may warn: choose More info → Run anyway.

Linux	File
AppImage (any distro, x64)	Muster-Agent-0.3.10-linux-x86_64.AppImage
Debian / Ubuntu	Muster-Agent-0.3.10-linux-amd64.deb
Archive	Muster-Agent-0.3.10-linux-x64.tar.gz

AppImage: chmod +x the file, then run it. Debian/Ubuntu: sudo apt install ./Muster-Agent-0.3.10-linux-amd64.deb.

Muster Server (self-hosted)	File
macOS, Apple silicon	muster-server-0.3.10-darwin-arm64.tar.xz
macOS, Intel	muster-server-0.3.10-darwin-x64.tar.xz
Linux x64	muster-server-0.3.10-linux-x64.tar.xz

Extract it, then bin/muster-server init and bin/muster-server start. It includes its own Node runtime. Docker, systemd and TLS setups are in docs/server.md.

Windows (installed with the setup.exe), AppImage and Debian/Ubuntu installs update themselves from Settings › Updates, downloading only what changed where they can.

Checksums for every file are in SHA256SUMS.

This build is not notarized by Apple (signing: self-signed). The first time, right-click Muster Agent in Applications and choose Open, or run:

xattr -dr com.apple.quarantine "/Applications/Muster Agent.app"

Muster Agent needs a model provider sign-in; it detects the ones already on your computer on first run.

Muster Agent 0.3.8

Muster now rests when nothing is happening. A finished chat had kept the graphics process and the window busy (about 29% and 13% CPU on a Mac), which warmed the laptop for no reason.

While idle, in a headless test	0.3.7	0.3.8
Frames drawn per second with a running-chat spinner visible	60.9	11.8
Same, window unfocused	60.9	2 (animations paused)
Compositor work per second (GPU thread)	29.5 ms	12.1 ms
Renderer script work per second	2.5 ms	0.11 ms
Screen updates per second with nothing changing	0.20	0

• Animations pause when the window is in the background. Spinners and pulses stop while Muster is unfocused or hidden and resume when you come back. The running-chat spinner now turns in 12 steps, like the macOS spinner, instead of redrawing the whole window 60 times a second.
• Clocks and polls rest too. "Up to date as of…" and sidebar ages refresh once a minute and only while the window is active. Pull-request checks, the hand-back countdown display and live-view timers pause in the background. The open-ports check backs off from every 4 s to every 30 s when nothing changes.
• The summary card no longer redraws itself every 15 seconds when the git status hasn't changed (#356).

Download

Mac	File
Apple silicon (M1 and later)	Muster-Agent-0.3.8-arm64.dmg (or .zip)
Intel	Muster-Agent-0.3.8-x64.dmg (or .zip)

Open the disk image and drag Muster Agent to Applications.

Windows	File
Windows 10/11 (x64)	Muster-Agent-0.3.8-win-x64-setup.exe (or the portable -win-x64.zip)

Run the installer; it installs for your user, no admin needed. It is not code-signed yet, so Windows SmartScreen may warn: choose More info → Run anyway.

Linux	File
AppImage (any distro, x64)	Muster-Agent-0.3.8-linux-x86_64.AppImage
Debian / Ubuntu	Muster-Agent-0.3.8-linux-amd64.deb
Archive	Muster-Agent-0.3.8-linux-x64.tar.gz

AppImage: chmod +x the file, then run it. Debian/Ubuntu: sudo apt install ./Muster-Agent-0.3.8-linux-amd64.deb.

Muster Server (self-hosted)	File
macOS, Apple silicon	muster-server-0.3.8-darwin-arm64.tar.xz
macOS, Intel	muster-server-0.3.8-darwin-x64.tar.xz
Linux x64	muster-server-0.3.8-linux-x64.tar.xz

Extract it, then bin/muster-server init and bin/muster-server start. It includes its own Node runtime. Docker, systemd and TLS setups are in docs/server.md.

Windows (installed with the setup.exe), AppImage and Debian/Ubuntu installs update themselves from Settings › Updates, downloading only what changed where they can.

Checksums for every file are in SHA256SUMS.

This build is not notarized by Apple (signing: self-signed). The first time, right-click Muster Agent in Applications and choose Open, or run:

xattr -dr com.apple.quarantine "/Applications/Muster Agent.app"

Muster Agent needs a model provider sign-in; it detects the ones already on your computer on first run.

Muster Agent 0.3.7

Work locally now hands back on its own, live diffs show up in checked-out tasks, router models no longer report "unavailable" after a restart, Ubuntu 24.04 installs start, and releases build about twice as fast.

Work locally hands back by itself, quickly (#345)

• No typing "done". After a finished turn Muster decides: a git project with a commit past the check-out (Muster commits uncommitted work as you) and a real test run that exited 0 after the last change (or no tests), or a plain folder whose files changed. Output Muster cannot parse no longer blocks; the exit code is the evidence. Errored or stopped turns, queued follow-ups and open questions never trigger it, and after Undo or Keep working it waits for new work.
• A 60 second countdown ("Handing back RAG-121 to QA Lead in 60 s", Hand back now, Keep working) runs in the app's runtime, so it survives a window reload. Undo still works afterwards. "Ask me" shows "Ready to hand back" instead.
• One PATCH, sent at once. Status, assignee and the summary go in a single request that jumps ahead of other waiting posts; the branch is pushed during the countdown.
• No "Paused" comment spam. At most one note per check-out, after a day of silence. "Not handing back yet" reasons now stay on the task in Muster.
• A correct summary. Files and lines come from the merge-base with the base branch, without gitignored or build output, plus uncommitted work. Unparsed tests with exit 0 say "Tests passed (exit 0)"; "read by you" appears only when you said done.
• Stale check-outs (idle over 3 days) show "Checked out 4 days ago · idle" with Hand back and Release in My work and on the task, and a plain-folder check-out in a very broad folder (home, Documents, Desktop, Downloads) suggests a project folder.

Live diffs in Work locally chats (#353)

• A checked-out task's chat now shows live inline diffs and the "N files changed" card, like any other chat, whatever the model, including edits made with shell commands.

Router models are there right after a restart (#350)

• Sending waits for a router's model list (for example intelligent-planner on Hybrow OmniRoute) instead of saying the model is unavailable. The last good list is remembered across restarts, failed listings retry on their own, and errors say exactly what is wrong.

Linux

• Ubuntu 24.04: the .deb now starts. It installs to /opt/muster-agent (a space in the old path broke Chromium's sandbox helper), and in containers without the capabilities that helper needs it falls back cleanly. The app's name in menus and its data folder are unchanged (#343).

Safer tools on router and local models (#348)

• Stop also ends programs the agent detached. Linked files outside the folder can't be read without Full access. Very large tool arguments are trimmed. Secrets are removed from stored commands and diffs. A failing project-rules check now blocks the command.
• The composer placeholder fits narrow windows, and its focus ring is clearly visible.

Faster releases (#341)

• Releases no longer re-run the full test suite: they check that main's CI passed for that exact commit. A version bump on main releases automatically once CI is green, downloads are cached, and the Windows zip uses normal compression. A release now takes about 5 to 8 minutes instead of 11.

Download

Mac	File
Apple silicon (M1 and later)	Muster-Agent-0.3.7-arm64.dmg (or .zip)
Intel	Muster-Agent-0.3.7-x64.dmg (or .zip)

Open the disk image and drag Muster Agent to Applications.

Windows	File
Windows 10/11 (x64)	Muster-Agent-0.3.7-win-x64-setup.exe (or the portable -win-x64.zip)

Run the installer; it installs for your user, no admin needed. It is not code-signed yet, so Windows SmartScreen may warn: choose More info → Run anyway.

Linux	File
AppImage (any distro, x64)	Muster-Agent-0.3.7-linux-x86_64.AppImage
Debian / Ubuntu	Muster-Agent-0.3.7-linux-amd64.deb
Archive	Muster-Agent-0.3.7-linux-x64.tar.gz

AppImage: chmod +x the file, then run it. Debian/Ubuntu: sudo apt install ./Muster-Agent-0.3.7-linux-amd64.deb.

Muster Server (self-hosted)	File
macOS, Apple silicon	muster-server-0.3.7-darwin-arm64.tar.xz
macOS, Intel	muster-server-0.3.7-darwin-x64.tar.xz
Linux x64	muster-server-0.3.7-linux-x64.tar.xz

Extract it, then bin/muster-server init and bin/muster-server start. It includes its own Node runtime. Docker, systemd and TLS setups are in docs/server.md.

Windows (installed with the setup.exe), AppImage and Debian/Ubuntu installs update themselves from Settings › Updates, downloading only what changed where they can.

Checksums for every file are in SHA256SUMS.

This build is not notarized by Apple (signing: self-signed). The first time, right-click Muster Agent in Applications and choose Open, or run:

xattr -dr com.apple.quarantine "/Applications/Muster Agent.app"

Muster Agent needs a model provider sign-in; it detects the ones already on your computer on first run.