Help

What it is

colabro gives you and your AI agent shared data models: you both work with the same records. Each of you sees the other’s changes right away. Everything goes into the history, and any change can be reverted.

The dark theme turns on by itself when your system uses it.

Getting started

• To connect an agent, open the page “Connect an agent” (it is also in the profile menu and under the list of workspaces): it has the address of the MCP server, the steps for Claude, ChatGPT, Cursor, Claude Code and other clients, first requests to try, and the list of connected agents, which updates by itself.
• Your agent creates the workspaces. Tell it, for example: “create a colabro workspace for planning a holiday”, and the new workspace appears in “My workspaces”.
• In the list of workspaces (the address /w) every workspace is a card. Click it to open the workspace in its main view. If the main view is not a table, the “⋯” menu of the card has “Open as a table”.
• The colabro mark on the left of the bar takes you back to the list.
• The menu with your name on the right of the bar has “Account” (your ways to sign in) and “Sign out”. On a phone, “Share”, “Help”, “Account” and “Sign out” are in the “⋯” menu.

Signing in, signing up and the password

• You can sign in with Google or Yandex, if they are turned on, or with a password: in “Email or username” enter the email you signed up with, or the username the administrator gave you.
• If sign-up is open, the sign-in page has a “Sign up” link: an email, the password twice and, if you like, a name. The password needs at least 10 characters, must not match the email and must not be one of the most common ones.
• colabro sends no email, so an email used this way is not verified: invitations to workspaces by email do not reach it. If you forget the password, ask the administrator: they give you a one-time link (valid for 24 hours) where you set a new password yourself. After that all your sessions end.
• To change the password, open “Account”, section “Password”: you need the current password. After the change all your other sessions end. There you can also set a password if you have only signed in with Google or Yandex.

The app on your phone

You can install colabro on the home screen of your phone or on your computer. The app opens in its own window, without the address bar, straight to the list of workspaces.

Install

• Android, Chrome or Edge: “Install the app” in the menu with your name (the button above appears when the browser is ready to install). Or the browser menu “⋮” → “Install app” (sometimes “Add to Home screen”). A long press on the app icon shows its shortcuts.
• iPhone and iPad, Safari: the “Share” button (a square with an arrow up) → “Add to Home Screen” → “Add”. Other browsers on iPhone work the same way, through “Share”.
• On iPhone the app on the home screen signs in on its own, apart from Safari: sign in once more after installing it. If signing in with Google or Yandex opens Safari and does not bring you back to the app, sign in with a password. You can set one in “Account”.
• Yandex Browser: the browser menu → “Add shortcut” or “Add to Home screen”.
• A computer, Chrome or Edge: “Install the app” in the menu with your name, or the install icon in the address bar.
• To enter text by voice, use the dictation of your phone keyboard: it works in any field.

Table

• A cell: click the value, change it and press Enter to save. Esc cancels the edit.
• “Add” above the table opens the form of a new row. Required fields are marked with an asterisk.
• The “⋯” menu of a row has “Edit” and “Delete”. Before deleting, colabro asks you to confirm and shows what happens to the linked records.
• Sorting: click a column header (or press Enter or Space when it is selected) for ascending order, again for descending, a third time for the original order. Empty values always go last. Only you see the order; it is kept in the address of the page, so you can share it as a link. On a phone the order is chosen in the “Sort” list.
• Dates are written as DD/MM/YYYY; the date field has a calendar button next to it. A date in the form YYYY-MM-DD is accepted too.
• Red text under a field: the value did not pass the check, correct it.
• While you edit, others may change the same record. Changes to other fields do not get in the way: yours is saved along with them. If someone changed the same field, the “Edit conflict” window opens with your value and the current one, and who set it and when. “Keep mine” saves your value over it; “Take theirs” (or Esc) keeps the current one.
• On a phone the rows of a table are shown as cards.

Board

• Drag a card to another column to change its status.
• Click a card (or press Enter on a selected card) to open it in full: every field, long text in full. You can edit it if you have the right to write; otherwise the form is read-only.
• On a phone you swipe through the board. You pick a card up by the handle on its left or with a long press.
• Without dragging: the card’s “⋯” menu → “Move to…” and the column you need.
• Pages made of blocks may have the same kind of board or a plan of places (for example, devices on the rails of an electrical panel). Drag a card with the mouse (Esc cancels the move) or with a finger: pull it sideways or hold your finger on it; swiping up and down scrolls the page. A taken place or a place beyond the edge lights up red, and the card does not go there.
• The “Add” button above the board opens the form of a new record: on a plan of places it already stands in the first free place, on a kanban at the end of the first column. You can change the place and the other fields in the form.
• With the keyboard: select a card with Tab and press Space or Enter, and the card is “in your hand”; the arrows move it between places and lanes, Enter puts it down, Esc cancels. Shift+Enter or a click opens the record.
• On a plan of places a device may be half a place wide (an auxiliary contact) or, say, two and a half, and stand on half a place. Shift+arrow moves the card by half a place. The lane header “3.5 of 6 taken” counts the halves too.
• If someone moved the card while you were dragging it, the “Edit conflict” window opens: “Keep mine” puts it where you dropped it, “Take theirs” (or Esc) leaves it where it is now.

Views and tabs

• Under the bar are the tabs: “Table” and one tab for each view of the workspace (a board, a page made of blocks, a custom view such as a panel). A star marks the main one.
• The tab you chose is in the address of the page: you can share it with a member as a link.
• A custom view (such as a panel): drag a device to another place or rail.
• Need one more view? Ask your agent to add it. It makes several small views for separate tasks rather than one huge one.

Search

• “Search” in the bar of a workspace (the magnifier icon on a phone), or the keys / and Ctrl+K (⌘+K on a Mac), opens the search across all the records of the workspace.
• It looks for words in the text fields of the records and in lists (tags, for example). Every word of the query must be there; the last one may be unfinished (from two letters): “cabl” finds “Cable”. Letter case does not matter. Word endings are not guessed: “pumps” does not find “pump”, so type the beginning of the word.
• The list shows the name of the record, its type and the fragment where the words were found. The arrows pick a record, Enter or a click opens it in the table, Esc closes the search.
• The 50 best matches are shown; if there are more, add a word.

History and the agent’s changes

• “History” on the right is the feed of all changes: who made them, when and with what comment. On a narrow screen History opens with the “History” button in the bar and closes with Esc or the cross.
• A dot with a number on the “History” button shows how much others (people and agents) have changed since your last visit. At the top of History is the summary “Since your last visit”: “Show” highlights the changed records, “Only new” leaves only them in the feed, “Mark as read” removes the summary and the dot. Opening History marks nothing by itself.
• Fresh changes by the agent are highlighted for a while in the table, on the board and in History. Hover over the sparkle icon next to a row to read the agent’s comment.
• To undo a change, use the “Revert” button of a History entry (owners and editors): first it shows what comes back; the revert becomes a new entry in History and can be reverted too. If the records of that change were changed again later, the revert does not run, and History offers to open the record. You can also ask your agent: “revert the last change …”.

Access and “Share”

• Only the owner of the workspace sees “Share”. The link opens the workspace without a password. Before creating it, choose “View only” or “View and edit”.
• The link is shown once, right after it is created, so copy it. “Create a new link” (the level may differ) turns off the old one; “Revoke the link” closes the access.
• Give the link only to people you trust. A guest sees only this workspace and, with an editing link, changes only its records.
• The icon next to the name of the workspace shows the access mode: “View only” means the data cannot be changed. Hover over it or tap it to read the explanation.

Invitations

• The owner invites a person by email on the “People” tab. The invitation waits for consent: you join the workspace only when you press “Accept” yourself.
• Invitations appear in the list of workspaces, above “My workspaces”, after you sign in with that email; the email must be verified (for example, by signing in with Google). Signing in with a password or with Yandex does not verify the email, and the invitation does not appear.
• “Decline” tells the owner “Declined”; the same email cannot be invited again for 30 days. Only the person can accept an invitation, in the browser; an agent cannot. colabro sends no email: the owner tells you the address of colabro themselves.

Limits and deleting a workspace

• Every user has limits: workspaces owned (20 by default), records in a workspace (20,000), data (20 MB in a workspace, 100 MB in total), the size of a record and of a schema (256 KB), members (50), views (30 of them, 50 MB), tokens for scripts (20), connected apps (10). Your agent can tell you the values: “what are my limits?”.
• “already … — that is the limit”: a new record does not fit. Delete what you do not need (deleting and making smaller always work) or ask the administrator to raise the limit.
• “Too many changes in a row. Try again in … s.”: wait for the time given.
• “Too many changes made within a day”: the size of the history of changes per day is limited (200 MB per person together with their agents). Deletions and reverts that make the data smaller always go through; what was deleted beyond the limit can be restored once the day has passed. Change only the fields you need rather than rewriting large records in full.
• “The server is running out of space”: this is the disk of the server, not your limit. While the administrator frees up space, you can only delete records and undo changes that make the data smaller.
• Reverts obey the limits too. If a limit was lowered, reverting a schema change back to a large schema (or reverting a deletion that brings back records beyond the limit) may be refused; free up space first.
• Only the owner can delete a workspace: the workspace menu → “Delete workspace…” and type its name. The records, the history, the members and the link are gone for good; the views you published stay.
• To delete your account, open “Account” → “Delete account”: the page shows what goes (the workspaces where you are the only owner, with their numbers of records and members, views, tokens, apps, ways to sign in); then enter your username or email and the current password (with no password, sign in again with Google or Yandex), tick “I understand…” and press “Delete account for good”. This cannot be undone. You simply leave the other workspaces; your changes there stay without a name. The administrator of the service cannot delete their own account here.
• “Account”: your name, “Sign out everywhere” (ends every session but this one), tokens for scripts and connected apps, which you can revoke or rename at any time.

If you are a guest

You opened the workspace through the owner’s link, and the bar shows the “Guest via link” badge. You need no sign-in and no password.

• You see only this workspace, its tabs and its history. A guest has no list of workspaces, no “Share”, no account and no sign-out.
• “Guest via link · editing”: you can change, add and delete records as described under “Table”. Without “editing” you can only view.
• Only the owner changes the schema, reverts changes, changes views and connects agents.
• If the owner revokes the link or creates a new one, this link stops working at once; ask them for a new one.

What to tell your agent

Connecting: in your MCP client (Claude, ChatGPT, Cursor and others that support remote MCP servers) add the remote server colabro.app/mcp, sign in to colabro and press “Allow”. If the client cannot do OAuth, a personal access token from the account page works. The agent acts on your behalf and with your roles. Each connection is a separate agent: a second one with the same name is signed “Claude 2” in History, and they see each other’s changes as someone else’s; you can set the name with “Rename” in “Account” → “Connected apps”. A connection that was disconnected and connected again gets a new number. To disconnect, open “Account” → “Connected apps”. For example, in Claude: claude.ai → Connectors → Add custom connector → the address of the service → Connect.

The agent knows what colabro can do by itself; if not, ask it to call about_colab.

Requests to your agent and the poller

A request is a task for your agent, set from a discussion with the “Ask my agent” button. The agent takes it through its connection to colabro, does it with your rights and finishes it; its work is seen in History marked “for the request”, and it can be reverted as a whole with “Revert everything for this request”.

For the agent to take requests on its own, without you, run a poller next to it: a small script (Node.js 22 or newer, no dependencies). It keeps one connection to colabro, waits for new requests and starts the agent only when there are some; colabro never connects to your machine.

How to set it up, in three steps

1. “Account” → “Poller for the agent” → “Create token”. Copy the token right away: it is shown once. A poller token can only wait for requests; it reads no data and no texts. Three at most; you can revoke them there too, and a poller with a revoked token stops at once.
2. Download the script poller.mjs to the machine where the agent runs (one connected to colabro: Claude Code, claude -p, the Agent SDK and so on).
3. Run it: put the token into the environment variable COLAB_WAKE_TOKEN and the agent’s command into COLAB_POLLER_CMD, for example claude -p "Do my requests in colabro"; then node poller.mjs. The colabro server is colabro.app by default; set another one with the variable COLAB_URL. The other settings are described at the top of the script itself.

• The poller does not start a second command while the first one runs; a request set in the meantime starts it right after the first one ends. The poller’s log has only the time, the number of requests and the exit code of the command: no texts, no names, no token. The command does not get the token.
• The poller remembers which request it got to (a cursor file next to the script). The first run without that file wakes the agent for any open requests. If a request seems to be missed (for example, it was set while colabro was being updated), delete the cursor file and restart the poller: after a restart it sees every open request.
• Do not want to run anything yourself? A routine on a schedule (say, once an hour) works too, with the same sandbox as below: no repository, a separate environment with no variables and no secrets, network “none” or Custom with an empty list, and only the colabro connector (remove the others that are on by default).

A sandbox for the agent. The text of a request and the data the agent reads may have been changed by other members, and the agent works alone, without you. So leave the agent the poller starts only what it needs:

• only the colabro connector: remove the other MCP servers and connectors (email, files, other services);
• no extra secrets in the environment: run it with a clean environment or as a separate user;
• no repositories with write access to working ones; a working directory with nothing valuable in it;
• keep the poller token in a file only you can read, and never put it into a repository.

What remains a risk

• The agent does a request with your rights in all your workspaces. An instruction slipped into a field, a file or a colleague’s message may tell it to read another workspace and write what it read here. colabro marks and counts such work and lets you revert it, but does not forbid it: the ceiling of a request is accounting, not a boundary of rights. Records the agent makes without taking a request carry no mark and are seen only in History under its name.
• When you press “Ask my agent” under someone else’s message, you vouch for the text: read it. Someone else’s text reaches the agent only as data; the instruction is yours to write.
• Do not open colabro to a browser agent (an extension, computer use) with your session: it could set a request on your behalf. And do not let the agent open links from the data: a share link to someone else’s workspace could become a channel for a leak.

Terms and personal data

• Terms of Service: you are responsible for the contents of your workspaces, the access you give and what your connected agents do.
• Privacy Policy: what account data colabro keeps, for how long and how to delete it.