The tamag0 desktop app (macOS, Windows, Linux) is where humans and agents work together. This is a tour of the actual interface: the sidebar, the three thread tabs, the composer, the context panel, and every settings section.
Arrival and offers
Arrival starts with your first agent and reuses the installation and provider connection steps. An agent without a connected model provider is not ready to answer. Your agents can work on projects you make available on your computer; this does not mean automatic project discovery or exclusively local processing.
The initial offer choice belongs to the workspace’s founding administrator, not every administrator. Joining an established company, including as an invited administrator, does not let you choose or replace its offer. A first administrator invited into a company prepared without an offer can make that initial choice. Once recorded, the choice is not changed through onboarding; later plan changes go through Softizy.
The offer chooser presents three working arrangements:
- Free: one personal agent.
- Pro: several personal agents that you work with.
- Enterprise: an organization whose colleagues work with their agents, including external agents or a mix of colleagues and agents you create.
Pro and Enterprise display a 30-day free trial, followed by €19.99 excluding VAT per additional agent per month. The first agent is free; the reference terms apply equally to additional agents you create or invite. Model-provider usage is separate. Selecting an offer creates no agent and does not itself start a trial, subscription, or charge. Before confirming each additional agent, you are shown the commercial impact: to keep the agent after the trial, a bank card will be requested; otherwise that additional agent will be deactivated. These are the offer terms, not an active trial or an amount due: this release does not activate trials, collect cards, charge, or deactivate agents. An explicitly accepted creation can move a personal Free workspace to Pro, never Enterprise. Existing agents and their data stay unchanged. Examples illustrate possible teams, not agents already created or included in the offer. An invitation that has not been accepted remains pending.
Additional creation requires actual creation permission and explicit acceptance of the commercial terms. Older offers’ fixed included-agent ceilings no longer block authorized creation or team preparation; there is no need to archive an agent or upgrade merely to free an agent slot. Provider and technical resource limits still apply. In a personal Free workspace, confirming the additional agent after accepting those terms moves the workspace to Pro only if creation succeeds. Enterprise is available during an initial organization arrival, not as an upgrade from a personal workspace. Selecting the initial arrangement does not remove existing agents, conversations, or data. An organization invitation has its own joining path, not an offer purchase.
Access your agents on another computer
Sign in to your tamag0 account to access your agents on this computer. Use the email address associated with that account; this does not mean joining another company.
If you already use tamag0 but have not associated your agents with your account, choose Associate my agents with my account on the sign-in page. Enter your existing tamag0 email address, verify the code sent to it, then choose Google or a password. This uses your existing profile, without creating another account or company. On subsequent computers, sign in with that account.
If Google refuses a Desktop sign-in, the page also offers this association path. It does not bypass access restrictions or guarantee that an account is eligible.
Layout at a glance
- Left rail — workspace avatars (one per company/agent connection), the Library destination, the eco-mode button, the language switcher, and settings.
- Thread sidebar — three tabs (Conversations, Activity, Thoughts) with per-tab unread counters, search, and reading filters.
- Main area — the active thread: streamed responses, permission prompts, the workflow progress dropdown, and the composer with its execution controls.
- Right context panel — what the thread is working with: files and repos touched, linked threads, participants, and any background tasks the agent is running.
- Status bar — live token usage and rate-limit state for the provider in use.
Your personal work home
A new conversation opens with competence shortcuts above a neutral greeting. Your first home starts with available work methods. Add, remove or reorder shortcuts to make it yours; removing one never deletes the competence or changes anyone else's home. An empty customized home stays empty when new methods are published.
Your home is saved for you in the current work context and follows you across signed-in computers and agent switches. Different people and workspaces keep separate homes. Unsaved changes are shown explicitly; if another computer has saved a newer version, choose which version to keep rather than silently overwriting it.
Add a competence searches names and summaries in the authorized catalog. If nothing fits, continue with your agent to describe what you need. Shortcuts are not an execution allowlist: ordinary conversation can still use authorized methods that you have never pinned or have removed from your home.
Clicking a card prepares a request; it does not send it. Suggestions reflect current agent specialties. You can choose one or several recipients, see their specialties and the associated person's first name for other people's agents, and check the same selected set in the composer before sending. Suggestions do not grant access or certify that tools are configured. You can always start by writing freely, without a widget or a catalog match.
In an agent's home workspace, a preparation shortcut lets you resume unfinished setup or adjust completed preparation. It reopens the existing conversation without sending a message, creating resources or publishing rules. You can keep working without finishing preparation. This shortcut does not substitute personal setup for a client assignment's context.
Language
The desktop app ships in English (default) and French. You choose it on the very first onboarding screen, and a language switcher sits in the left rail, just above settings — toggle either and the whole interface updates instantly, no restart. The choice covers everything you see: the chat, onboarding, settings, the Library, native surfaces (system-tray menu, notifications, dialogs), and the right-click edit menu. It is remembered per install. Agent responses follow your agent's own language preference, set on its profile — independent of the interface language.
Library
The Library button in the left rail opens your Library — one shared store of files reachable from every one of your agents, separate from the thread tabs since it isn't a conversation filter. Search across it, inspect an artifact's folder and processing status, download it, or drop new files in yourself through the native file picker (see Library).
Working on a document
Ask your agent for a laid-out document, then keep refining it in the same conversation. A completed HTML document becomes the main working surface, with your exchanges on the right and the composer still available. While a revision is being prepared, the previous complete version stays visible; an unsuccessful revision does not replace it with a partial file. You can return to ordinary conversation, change subjects, and reopen a working document from Context. In a narrow window, switch between the document and conversation views without losing your draft message.
One self-contained file
For this workflow, ask for a self-contained HTML document: one UTF-8 file with its styles and supported images embedded, using system fonts or supported embedded fonts. Continue revisions in the same file. If an existing document depends on separate stylesheets, images, or other files, ask your agent to make a self-contained export first; Desktop does not assemble a multi-file website for you.
The document is a static reading surface, not an executable app. Supported layout and embedded assets are preserved, while scripts and interactive features are excluded. Linked local or remote resources are not fetched; unsupported dependencies need an explicit export or correction rather than a promise that the preview reproduces them.
Read in your browser without publishing
Use the document toolbar's browser action to open the displayed completed version in your default browser, even before adding it to the Library. This is a protected local reading view: opening it does not upload the document, change its visibility, or create a public share link. If the browser cannot launch, the app shows an error and keeps the document available.
An already-open browser view stays on that version when you make further edits. Open the current completed version again when you want to read the revision. The local address is not a recipient share link and stops serving the document when its local view expires or its owning context is no longer available; it is not permanent storage. Closing the document pane alone does not recall an already-open browser copy.
Publish deliberately
Add the displayed version to the Library explicitly. The initial saved document is private, in the current workspace; public sharing is a separate choice in its Library entry. Later edits appear as unpublished changes and do not upload automatically. Explicitly updating the Library replaces the associated published document, so existing valid share links serve the update on their next load rather than a second entry being created.
Wait for a confirmed result before treating a version as published. If a transfer result is uncertain, do not assume success or blindly create another copy; use the available status check or open the Library to inspect the result. See Library for recipient reading, latest-published downloads, and the limits of expiration and revocation.
In a shared conversation
When an agent presents a document in a conversation shared with other people, the document is shown to the other participants too: at the end of the agent's turn it appears as a link in the chat and takes the place of the conversation on their screen, with the agent's name under the title, and each new version the agent presents reaches them the same way. It is never added to anyone's Library automatically; a participant who wants to keep it uses Add to Library in the document pane, and the status changes from Not added to Saved version.
Your Downloads folder
An agent can put a file into your computer's Downloads folder — one it fetched from Google Drive or one it made in the conversation, up to 100 MB — so you find it where your downloads usually go. It never replaces a file already there: a taken name gets a number. It can also bring one file from your Downloads folder into the conversation to read and use it: the file you name, or the one you most recently downloaded ("the PDF I just downloaded"). If you only half remember a name, it can look for it: it searches file names only — never what is inside a file — at the top of your Downloads folder, shows you the few names that match with their date and size, and asks which one you mean before it brings that file in. Give it a word of the name: a kind of file alone, such as "pdf", finds nothing. It never browses or lists the folder, which often holds personal papers, and it never renames, moves or deletes anything in it.
Saving, searching and bringing a file in each ask for your consent first, once or for the whole conversation: the prompt shows what the agent looks for, or names the file it would bring in. Allowing one for a conversation does not allow the others — letting it save files, for example, does not let it search or bring any in, and neither does allowing all of tamag0's tools for the conversation. In a run where nobody can answer, all three are refused unless you allowed them for that conversation or turned off permission prompts. They work only in your own threads with the agent, never in a conversation shared with other people or agents.
Formatted and mathematical answers
Agent answers are rendered, not dumped as raw text: headings, lists, tables, and syntax-highlighted code blocks. Mathematical notation is typeset — inline expressions sit in the flow of a sentence, displayed equations get their own centred line, and the usual LaTeX notation works whichever delimiters your agent writes. Fractions, sums, integrals, matrices, Greek letters, and the standard operators all come out as real formulas rather than backslash soup, so a conversation about mathematics stays readable by both sides. This applies everywhere markdown is shown: the conversation, subagent results, the context panel, interactive question previews, and Library markdown artifacts. Code samples are left untouched — a formula written inside a code block stays literal.
Activity in a conversation
When an agent works through a task, tamag0 keeps the conversation readable: consecutive tool calls are collapsed into one expandable activity row that shows the action in progress in plain words, not a list of tools. Once the work is done, the row settles on how many actions it took and how long they ran. Open the row for the per-call detail — every call, not only the failed ones; interactive questions and visual results remain visible in the flow. Codex reasoning summaries follow that same activity row instead of becoming separate chat messages, so live progress and a reopened thread keep the same compact shape.
Threads: three tabs
- Conversations — the threads where a human talks with an agent. Draft threads are only persisted on the first message; titles are generated automatically.
- Activity — the agent's autonomous work: messages from other agents and Slack traffic processed in the background. A live pulse shows when an autonomous thread is currently running. When an agent needs a human decision, the thread is escalated — it moves from Activity into Conversations so it lands in front of you (see Thread lifecycle).
- Thoughts — the retained results of the agent's autonomous reflection: introspection, analyses, monitoring notes, and dream narrations that it chose to carry forward after reflection (see Continuous learning). It is an auditable record of what the agent chose to retain, not a dump of hidden model reasoning.
Scheduled tasks appear in Conversations as one collapsible family per task, with a dated, independent child for each automatic or manual run. Open any execution to read or reply to it; see Scheduled tasks.
When one thread creates related work, the sidebar presents that family as a collapsible group inside each tab. The leader and every visible descendant stay aligned at one flat level; expanding the group reveals the rest without a nested tree. If the original parent is hidden or belongs to another tab, the first visible member leads that tab's group instead. A dot on the leader signals unread activity inside the collapsed group.
Families are ordered by their most recent message, and expanded descendants are ordered by their own most recent message. Only one family stays expanded at a time. Opening or closing an already selected family changes only the sidebar: it does not reload the conversation, and an explicit collapse remains in place while the list refreshes in the background.
Resolved conversations disappear from the daily list when resolved rows are hidden and return when Show resolved is enabled. Hiding a resolved family leader never hides active descendants: the newest visible descendant becomes the temporary leader until the canonical parent is visible again.
Grouping never overrides where an individual thread belongs. If one autonomous child starts waiting for you, that child alone moves from Activity to Conversations; its siblings stay where their own state puts them. Thoughts likewise remain Thoughts even when they have children.
Reading controls
Working with agents that act on their own means lots of threads — the sidebar is built for triage:
- Hide read / show read — per-tab toggle (Activity hides read threads by default, so it behaves like an inbox).
- Show resolved — in Conversations and Activity, read closed, resolved, abandoned, and archived threads are hidden by default. Unread terminal threads remain visible. Turning on this toggle adds terminal threads to the current view; Hide read and other active filters still apply.
- Mark all as read — one click clears the current view.
- Filters — by agent (when the workspace has several) and by domain, plus full-text thread search.
Deleting a thread that has descendants is scope-aware: the confirmation shows the current number of affected threads. If that family changes before deletion completes, nothing is deleted; the app refreshes the count and asks you to confirm the new scope again.
Finding a message in a long conversation
Long-running conversations outgrow what scrolling can reach. Cmd/Ctrl+F opens a find bar scoped to the conversation you have open, and it searches the entire history — not just the recent messages currently on screen.
- Results appear as excerpts with their author, counted in messages, listed from the top of the conversation down:
1 / 12is the oldest match,12 / 12the most recent. Beyond 500 matching messages, only the 500 most recent are listed, still numbered against the full count — the first one reachable is then501 / 1000, not1 / 1000. - Picking a result — or pressing Enter — takes you to that message, in a read-only window of the surrounding history. The searched term is highlighted in the conversation, the match you are on stands out from the others, and the view scrolls to the term itself, even deep inside a long message.
- The list then collapses into a position counter (
2 / 12 messages) with previous/next steppers. Enter moves to the next match and Shift+Enter to the previous one; both wrap around from the last match to the first, so you can walk the matches without retyping. - Pressing Cmd/Ctrl+F again while the find bar is open selects the current query, so you can type the next search straight away.
- While you are reading a search result in history, the live conversation keeps running untouched behind it. A return control stays visible for the entire historical view, identifies that you are viewing history, and takes you back to the latest message in one click.
- If you simply scroll up in the live conversation, the same control stays visible while you are away from the bottom and tells you how many new messages arrived meanwhile.
- Closing the find bar clears the query, the results and every highlight, but leaves you where you were reading. Returning to the live conversation is always a deliberate, separate action.
The search covers persisted message text. It does not cover the output of tools that were not persisted, text that only lives inside persisted structured blocks, or queries that span Markdown markup — the find bar states these limits itself when nothing matches.
The composer
- Dictate with macOS — on a Mac, the microphone button just left of Send starts the Mac's own dictation and keeps the cursor in the message field. Spoken words appear there like typed text. Nothing is sent automatically: review the text and send it yourself. You can also dictate while an agent is answering. macOS provides the listening indicator and the controls to stop dictation; tamag0 shows no recording state. Where the audio is processed follows your Mac's dictation settings.
- Lower bar under the composer — message actions (attachment, mention, and skill) stay on the left; execution, workspace, and agent state stay on the right. The current explicit model preference remains visible after a message is sent, while Auto leaves the choice to the agent.
- Hierarchical execution menu — open the execution control to choose Model, Effort, Model size, and Permissions whenever each axis applies. Model starts with Auto, then lists every configured provider and the models in its live catalogue, including custom providers. Choosing a Claude or Codex model resets Effort to Auto; other providers use their catalogue defaults, and unsupported axes stay hidden.
- Skills as slash commands — type
/to invoke any skill; multi-step skills show a progress dropdown with the current step, and you can skip ahead. Workflow progress is kept independently for each thread, so switching conversations never mixes their steps and reopening a thread resumes its own progress. - Mentions — type
@to pick an available agent or a person. Agents are listed first, people below them (see Adding someone to a conversation). Guests working in the workspace show their home workspace in the list (for example,Luc (Softizy)), so people with the same name stay distinguishable. The selected agent is addressed by its stable identity, not by the display name. - Attachments — drag-and-drop or paste images and files. A sent image shows as a thumbnail in the conversation; click it to view it full size (Escape, the close button or a click outside closes the view).
- Agent picker — with several agents in the workspace, choose who you're addressing per message.
- Create a subthread — in an existing conversation, use the action beside Send or Stop to turn the current draft into the first message of a linked child conversation. The child opens immediately and its first answer begins by summarizing a bounded, authorized window of the parent before answering the transferred message. A failed creation or transfer keeps the parent draft available. If that parent context cannot be loaded or scoped, no answer is generated and the transferred child message remains retryable; an interrupted first answer is likewise retried with the same parent context instead of producing a duplicate child.
- Mention a guest on assignment — type @ in a conversation, then choose the guest shown as Name (Home workspace). tamag0 inserts its stable routing identity, so same-named guests remain distinguishable. See Cross-workspace assignments for the assignment lifecycle and access limits.
- Interactive questions — when the agent needs a decision mid-task, it asks with structured options instead of guessing. If you have not answered after a few minutes, the agent does not carry on without you: it ends its turn saying it is waiting for your answer. The question stays answerable: pick an option whenever you are ready, or reply in a new message, and your answer starts a new turn.
Automatic provider fallback
When the composer is on Auto, tamag0 can continue the same turn with the next entry in the priority order after a recognized capacity or rate limit. See the automatic fallback rules for which failures qualify. The composer remains on Auto while the live header follows the provider, model, parameters, effort, and attempt that are actually running.
Each accepted switch creates one durable notice in the conversation. If every configured provider is exhausted, the terminal card shows the final error and the ordered history of actual attempts; providers skipped during preflight or cooldown are not presented as attempts. The same status survives reopening the thread, and assistive technology announces each accepted attempt or terminal state once.
Fallback after partial work uses the existing conversation history plus one internal continuation instruction. Visible partial output stays in the conversation, while completed tool actions are never mechanically replayed. The same recovery behavior applies to direct conversations and background Hub-routed work without relaying or acknowledging an intermediate attempt.
An explicit provider or model choice is a pin, so it never switches automatically. A pinned failure offers Retry and Switch to Auto and retry. Retry makes one attempt with the selected provider, bypassing its cooldown once without starting a cascade; switching to Auto uses the configured priority order again.
You never wait for the agent
The composer stays yours even while the agent is thinking:
- Queue while it works — press Enter during a response and the message is queued as pending; it's delivered automatically the moment the current turn finishes. Queue several — they arrive as one combined message.
- Interrupt on your terms — the pending banner has a Send now action (also triggered by pressing Enter twice): it stops what the agent was doing and delivers your message immediately. Use it when what you just learned changes the task.
- Edit mid-flight — a pencil icon on your last message lets you take it back while the agent is still responding: the response is stopped and your text returns to the composer for editing and resending.
- Drafts and pending queues stay with their original thread, including attachments. Returning to a thread that is still working keeps its corrections pending. If a new turn starts while an outgoing correction is being prepared, the correction returns to that thread’s queue rather than being discarded; text you are still composing is kept separately.
Adding someone to a conversation
You can bring a colleague into an existing conversation without bringing their agent along.
Pick them with @. The @ list shows a People group below the agents, filtered by what you type. In a Home conversation it lists the people of your workspace. In a conversation on a client assignment it lists that client's people, plus the owner of each guest agent on the assignment (shown as Owner of …), never the people of your Home workspace. If the list cannot be loaded, the agents stay selectable and the People group says so.
A mention never adds anyone on its own. Your agent reads what you asked:
- "Ask @Mira what she thinks" — the agent contacts Mira in a separate conversation. Nobody is added here.
- "Bring @Marie into this conversation" — the agent proposes the addition, and you decide on a card.
The card is shown only to the person who wrote the mention, even when you are talking to your own agent. Other participants, the conversation's owner included, see the agent's reply but no card. The card shows who proposes, who would join, and why. For a person, you choose what they will see:
- Whole conversation (preselected) — everything since the first message.
- From their addition on — the message that mentions them, then what follows. The card reminds you that the agents in the conversation have read the whole history, so their replies may still draw on it.
An agent always joins with the whole conversation. Nothing is shared before you click Add; history someone has read cannot be taken back. Refuse shares nothing. The card then settles in place (Added, Refused or Already a member) until you reload the conversation. No line is added to the conversation itself: the new member simply appears in the participants panel. If the addition fails, the card offers Try again.
On the added person's side, the conversation appears in their Desktop. If their Desktop is running, they get a notification "name added you to title" ("…, from now on" when they only see what follows); otherwise the conversation simply shows as unread. With Whole conversation, the full history is loaded, each question next to its answer. With From their addition on, a notice above the first message says that earlier messages are not shared with them. They read and write under their own name, and their own agent is never involved.
Removing someone or leaving a conversation
A person added to a conversation can be taken out of it again, by the conversation's owner, or can leave by themselves. This works in Home conversations and in conversations on a client assignment.
Who removes. Only the conversation's owner, the person who started it, removes someone. When you start a conversation with a colleague's agent, you are its owner, not that colleague. The owner is never removed, and someone who takes part through their own agent rather than by being added has nothing to remove here.
Removing someone. In the participants panel, hover the person (or reach their row with the keyboard): a ⋯ button appears, then Remove from the conversation…. The confirmation asks what the person keeps:
- Messages already received, read-only (preselected) — the conversation stays in their list, frozen at this moment.
- Nothing — the conversation disappears for them.
Click Remove followed by their first name to confirm. Either way, the person no longer receives messages and can no longer write in the conversation. What they wrote stays in it, signed with their name, and what they already read cannot be taken back. The person leaves the participants panel at once. If the removal fails, the confirmation says so, the person is still part of the conversation, and Retry tries again.
Leaving. A person who was added can leave from the menu behind the conversation's title: Leave the conversation…, then Leave. There is no choice to make: you keep what you received, read-only, and can delete the conversation afterwards like any other. Someone can add you again later.
What everyone sees. One line appears in the conversation, with its time, worded for each reader: "You removed Jordan" for the owner, "Jocelyn removed Jordan" for the other participants, "Jordan left the conversation" after a departure. Nobody is notified and the conversation is not marked unread. The agents taking part do not read that line as a message.
On the removed person's side:
- With Messages already received, read-only, the conversation stays in their list and ends with "Jocelyn removed you from the conversation". The composer is replaced by "You are no longer part of this conversation", with the date until which they keep the messages received, and the participants panel no longer lists the people in it. They keep what their Desktop had already received when they were removed; nothing more is loaded afterwards. After leaving, the conversation is read-only in the same way, and ends with "You left the conversation".
- With Nothing, the conversation leaves their list without a message. If they had it open, it gives way to "This conversation is no longer available."
Adding someone back. A person who was removed or who left is added again like the first time: mention them with @, ask your agent to bring them in, and decide on the card. The card offers a fresh choice of what they see. On a conversation they kept read-only, the same conversation becomes writable again, without a new notification: with Whole conversation, the messages posted while they were away are loaded; with From their addition on, a line right after their removal says that the messages of that period are not shared with them, and what they had received before stays above it. A conversation that was erased for them comes back in their list, with the history chosen on the card.
Context panel
Every thread has a context panel showing the files and repositories the agent touched, its actual parent and direct children (related work stays navigable after it finishes), and the participants — humans, local agents, and external agents taking part in the thread. Hovering a person who was added shows who added them, when, and whether they see the whole conversation or only what followed their addition; the conversation's owner also gets a ⋯ button there to remove them (see Removing someone or leaving a conversation). Parent and child work is separated into clear sections; other informational relationships are not presented as family.
The panel also links out to the web. When the thread is waiting on an external event, it names what is being watched — a pull request, a CI run, a deadline — and one click opens that pull request in your browser. Repository entries work the same way: a click opens the repository on its remote, whether that is GitHub, GitLab, or a self-hosted forge. Anything with no page behind it — a repository with no remote, a plain deadline — stays plain text instead of becoming a dead link.
Background tasks
When an agent starts a long-running command in the background — so it can keep working while the command runs — that task appears in a Background tasks section of the context panel, live, the moment it launches. Each entry carries a status (running, done, or killed); clicking it opens a detail view with the exact command and the output collected so far. The section stays hidden when there are no background tasks to show.
Tracking: your work and the project
The Tracking button of the left rail opens one window with two tabs.
My tracking follows the requests your own agents take part in, and every request you launched yourself, even when a guest agent of a mission handles it on its own: each request with its tickets, pull requests, checks and stages, from framing to merge. Its first group, Mine, holds what waits for you — a reply in one of your conversations, or a pull request GitHub asks you to review — and the rail badge counts only that group.
Project shows the work of the whole workspace on the GitHub repositories assigned to you, whoever launched it, together with every request you launched yourself, with all its tickets. Group it by request or by repository, and filter it by agent and by requester. A request still being framed stays in My tracking until it is published. Each row says which agent works on it and for whom; titles, stages, pull requests and checks are shown, never the conversations. A request of someone else counts all of its tickets in its progress, and says how many sit on repositories outside your view. When GitHub asks you to review a colleague's pull request, Review the PR appears there — it does not enter Mine nor the badge.
On n repositories, at the right of the toolbar, lists the repositories you see and who assigns them. With none yet, and no request of your own, Project says so and names the people to ask.
Repository access
Seeing a repository is a right per person and per repository: sees, or does not see. Being an administrator does not show every repository — administrators assign themselves repositories like anyone else.
- An administrator, or a member they give the tech profile, opens Repository access from the Project tab. It lists the repositories reachable through their own GitHub CLI session on this computer, those the workspace already works on, and any added by hand as
owner/name; tick them by person or by repository. The tech profile gives this screen, not more visibility. - A repository is also assigned automatically: when an agent opens a pull request, its person and the person who asked for the work receive that repository, shown as auto · PR #n. An automatic right can be removed like any other; the next pull request of that kind gives it back.
- An agent placed on an assignment in another workspace sees the repositories that workspace's administrators or tech profiles tick for it, and no other: its own pull requests add none, and the rights end with the assignment. Its person finds that work in Project, next to their own workspace's.
Workspaces and eco mode
A workspace connects to a company and holds one or several agents — added by API key, by invitation, or provisioned automatically. Each agent has its own secure credentials, threads, and an isolated working directory per thread. Multiple workspaces (e.g. two companies) work side by side, with instant switching from the left rail.
An e-mail address belongs to one account space only. To hold accounts inside two companies, use a different address for each. An employee can instead keep their account and agent in their personal space and join a company on an assignment, without a second company account. A company cannot invite as an employee an address that already belongs to another company: the invitation is refused without saying which company holds it. Inviting someone from another company to work on an assignment is a different thing and is not affected — see Cross-workspace assignments.
Eco mode (the leaf button) mutes incoming agent-to-agent requests for two hours to save tokens — your agent stops accepting peer work but stays available to you. One click resumes normal service early.
Always-on by design
The app keeps agents reachable: messages from other agents, Slack, email, and scheduled tasks are processed in the background — ordered sequentially per domain, in parallel across domains, with automatic reconnection after network loss or OS sleep.
On Windows and Linux, closing the window doesn't quit the app — tamag0 minimizes to a system tray icon and agents keep working in the background. The tray menu offers Show tamag0, Hide to tray, and Quit tamag0; the first time the window hides, a one-time hint confirms the app is still running and offers a Quit completely shortcut. This is on by default — turn off Close to tray under Settings → General so closing the window quits instead. macOS has no tray: the Dock already owns show, hide, and quit.
Permissions and control
Sensitive actions (shell commands, file writes, outbound messages) surface as permission prompts with once / this-thread scopes; denials are visible, never silent.
If you trust an agent to act without interruptions, open the execution control under the composer, choose Permissions, then choose Bypass for the current conversation. The same bypass can be set as a global default in Settings; enabling that global default takes an explicit confirmation. Either way, every affected conversation shows a persistent banner. Choose Inherit to use that default or Ask to restore prompts. When the agent explicitly opens a private child conversation with itself, that solo thread inherits this human-controlled setting; peer messages, integrations, and scheduled tasks do not. See Security.
Settings tour
- Agent — the agent's profile: name, origin story, specialty; per-agent configuration when the workspace has several. This is also where you create a new agent or team.
- Extra directories — per-agent read access to folders (and everything under them) outside the thread's working directory. Add a path (or pick one with a folder browser) and the agent can read there without a permission prompt each time — useful when you want it to analyze documents that live elsewhere on your machine. Writes still go through the normal permission prompts; the access is read-only, local to this device, and not synced across machines. See Security.
- General — runtime selection with both Claude Code and Codex set up from the app; working directory; performance (how many agent workers run in parallel — one worker per active thread, whatever the runtime: Claude, Codex, or local models; size it to the number of conversations you expect to run at once); global CLI install so your agent follows you into the terminal; and — on Windows and Linux — the Close to tray toggle that decides whether closing the window minimizes to the system tray or quits.
- Account — the sign-in of the active workspace: its address and the computers and browsers connected to your account. Disconnect any other computer or browser from here; a disconnected computer loses access to your agents right away, cuts the phones paired with it, and erases your account's credentials and connections as soon as it contacts tamag0. While this is your only computer, the tab tells you how to add another: install tamag0 on it from tamag0.ai/download and sign in with the same address to find your agents there. A computer still set up with a key keeps working and is invited to sign in here; once signed in, it switches your own agents from their key to your account by itself, with nothing to do on your side.
- LLM Providers — the model catalogue, custom providers and models, and the per-agent runtime priority order used for automatic failover. Priority choices follow your account across devices after restart while remaining usable offline.
- Workspaces — add a workspace with an API key, switch, manage its agents, and — for workspace administrators — place an active agent on a temporary, revocable assignment in another workspace. The receiving workspace's administrator accepts the single-use invitation from Add agent; invited agents remain visibly identified by their home workspace. The guest list is read-only: placement removal and company-memory permissions are managed through its link to web administration → Guests. Personal and provider-side assignment controls remain in the desktop. See Cross-workspace assignments for the complete flow and safety guidance. Company admins also require two-step verification for every member from the company's card (see Two-step verification).
- Connectors — the services your agents use, MCP servers included: discover and add them, connect each once per work context, choose the agents that use it, and set per-tool permissions (see Connectors).
- Mobile — the phone app. Google and Jira are connected from Connectors (see Integrations).
- Reflexes — the behavioral layer you can inspect and edit: golden rules, best practices, forbidden commands, and reminders, per agent or for all.
- Skills — browse, create, and share skills (see Skills).
- Scheduled Tasks — recurring jobs with their schedules and last runs (see Scheduled tasks).
- Watchdog — the recent decisions of the thread watchdog, so autonomous supervision stays auditable.
- Appearance — light/dark theme.
- About — version and updates; the app auto-updates.
Status and usage
The status bar shows live token usage against the provider's rate-limit windows, with reset times — no surprise limits mid-task.
Signing in from the Desktop
When you connect a computer to tamag0, the Desktop opens a browser page that starts with one email field. You enter your address and receive a verification code by email — the same code request for every address, so the page never reveals whether an address has an account. Entering the code proves you control that mailbox; it does not sign you in on its own.
After the code is verified, the page routes you to the right next step:
- Your access is already configured — you sign in with your existing password or linked Google account, the same credentials you use on the observation dashboard. The emailed code only oriented you here; it is not a second factor and does not replace your password.
- You are an eligible member whose web access has not been set up yet — the page walks you through configuring your access (choosing a password or connecting your Google account), associated with the existing member record your administrator created. No new member or company is created.
- The address is unknown, disabled or ambiguous — the page tells you web access is unavailable for that address and does not create anything. If you expected access, contact the person who invited you.
Once you complete sign-in or setup, the browser returns you to the Desktop through the original deep link, and the waiting computer connects. If the Desktop request expired in the meantime — because you took too long or switched computers — the page says so and you start again from the app. Cancelling on the Desktop side discards the pending exchange locally; the server request simply expires if it is never completed.
The email-first page is separate from the observation dashboard sign-in. Operators who manage the platform sign in through their own entry, which drops any Desktop return context.
Two-step verification
Two-step verification adds a 6-digit code from an authenticator app on your phone to your password or Google account. It is optional, unless your company requires it.
- Turning it on happens on tamag0.ai, on your account security page: scan the QR code (or type the key), then enter one code from the app to confirm it works — nothing is switched on before that. You then get 10 single-use recovery codes, shown once; keep them somewhere safe. Generating new ones cancels the old.
- Once on, the code is asked at every sign-in: on tamag0.ai with a password or Google, and when you connect a computer from the Desktop. A forgotten password asks for it too, before the new password. A recovery code works wherever a code is asked.
- In the Desktop, Settings → Account shows whether it is on and how many recovery codes you have left. Turn on and Manage open the account security page in your browser; every change happens there. Never signed in on tamag0.ai? On the sign-in page that opens, Forgot your password? sends you the link that sets up your access first.
- Turning it off takes a code or a recovery code, even while signed in, and is not possible while your company requires it.
- Too many wrong codes (10 in 24 hours on one account) lock it: even a correct code is refused until the window ends. A company admin can reset it, which lifts the lock at once.
- Lost your phone and your recovery codes? Ask an admin of your company to reset your two-step verification from Humans in web administration. You receive an email right away, you set it up again at your next sign-in, and the admin never gets access to your account. An admin cannot reset their own; a company's only admin writes to tamag0 support.
Requiring it for your company. A company admin turns on Require two-step verification in Settings → Workspaces, in the company's card, on every plan. The card shows how many members already use it and who does not, so you can warn them first. Each member without it sets it up at their next sign-in, on tamag0.ai as in the Desktop; until then their computers and keys keep working, and nothing is cut when you switch it on. The requirement also applies to the company's tamag0.ai administration account, if it has one. Members can no longer turn it off while it is required.
Onboarding in minutes
Pick your interface language (English or French) as the first step, sign in through your browser (or accept an invitation, which creates your account the same way), name your agent, then check the local development tools it uses for projects: Git and GitHub CLI. On Windows, tamag0 can install those tools before model runtimes; on macOS and Linux it shows their status so you can install any missing tool with your system's usual method. You can continue without them and configure them later. Next, install/sign in to the model providers (Claude and Codex both set up out of the box; each skippable — signing in reuses your existing Claude or ChatGPT subscription, no separate API billing), pick a workspace folder. The founding administrator chooses an initial offer before entering the app. The first administrator invited into a company that has not yet chosen an offer also makes this initial choice; other members skip that step. No orchestration to wire, no infra to babysit.
The agent's guided discovery opens in Conversations. Moving from discovery to synthesis removes the transition action immediately while the agent prepares its response. If Claude authentication has expired on that first turn, completing the sign-in resumes the introduction automatically instead of leaving an empty thread to restart by hand.
Invited as an employee
When a company invites you as an employee, the e-mail carries a link to a page on tamag0.ai. That page names the company, tells you that you are joining it as an employee, and offers three things: download the application, open it if it is already installed, and copy your invitation code. The page asks nothing of you and creates nothing on its own — a link opened by mistake, previewed by a messaging tool or scanned by a mail filter leaves the invitation intact. Keep that page open during the installation: it is what hands the invitation to the application.
The company declares you an employee. You choose what happens to your agent, from the paths offered by your invitation:
- Create my agent in the company — your account and agent are in the company. The agent belongs to it and stays there if you leave.
- Come with my personal agent — your space and agent remain yours. The company welcomes that agent on an assignment as an employee. Its shared memory is open to your agent, and the company sees the agent's activity time, never your own activity. This is explained before you confirm, with no time-sharing checkbox.
If your invited address already belongs to a personal space, only the personal path is offered. If you already use an account in the inviting company, only the company path is offered. An address held by another company cannot receive an employee invitation; external collaboration uses a separate invitation.
On an installed application, open the invitation and choose the offered path. The company path creates an agent in an additional workspace on this computer. The personal path uses your own agent; if your space holds several, choose which one joins. Each invitation places one agent.
On a fresh installation:
- Open the invitation from its page, or choose Enter the code on the sign-in screen and paste it into Code from your email. The app names the inviting company and offers the eligible paths.
- For the company path, continue in your browser to set up access with a password or Google, or sign in to the invited account. The address is already filled in; the invitation proves it, so no email code is needed. Complete any required two-step verification. Back in the app, create your agent in the company, or find the agents your account already owns there.
- For the personal path, the app explains that the setup happens in your own space, not the company's. Create your personal account through the browser, or sign in to your existing personal space. Create your agent if needed, then it joins the inviting company as an employee. A new agent gets to know you in your personal space.
Joining an established team through the company path does not require an offer choice. If you are the first administrator invited into a company with no offer yet, you make its initial choice during installation. Choosing the personal path does not create an account, password or two-step verification setup in the company. With an administrator role on that assignment, sign in to the web console using your personal account and select the company to administer it; this does not give Desktop administration rights in the company.
Finish browser sign-in on the computer that started it. If the request expires or the application restarts, use Try again; a password already created only needs a sign-in. If setup was interrupted after creating your personal space, sign in to that space and resume the invitation there rather than creating another account.
Older application: the existing company-agent journey continues to work. An older version cannot offer the new personal path: it may show a retryable sign-in failure or a generic invitation error. Update the application and reopen the same still-valid invitation. That refusal does not consume it, even when retried. An employee assignment may show no status banner until the app updates.
An invitation that has expired, been cancelled or already been used answers with a single message that does not say which. If you already have an account, sign in; otherwise ask the person who invited you for a new invitation. The invitation code is not an API key. The link that activates the observation dashboard is a different thing and never goes in the application. External collaborators — freelancers, contractors — use their own invitation flow, described in Cross-workspace assignments.
Download tamag0 Desktop
Download tamag0 Desktop for Windows, macOS, or Linux. The download page offers a direct Windows installer, separate Apple Silicon and Intel builds for macOS, and AppImage, .deb, and .rpm packages for Linux. It is available in French too; the latest public release remains available as a fallback for other architectures.