A shared board your whole team writes on at once, walked through card by card, with the actions and agreements it produced still there at the next retrospective. This page goes through every feature, in the order a retrospective happens.
A retrospective belongs to a team rather than to whoever called the meeting, so everything starts with one.
From Your teams, choose Create team, give it a name and a colour, and you are its owner. You can hold as many teams as you like and switch between them from the top bar.
A team manager invites several people at once, by email address, and picks the role they arrive with.
The roles are about managing the team, never about running a meeting: every member writes, reveals, presents and finishes a retrospective alike.
A private team is readable by its members alone. A public one can also be opened by anybody signed in here, as a guest. Each retrospective can be made private on its own, which keeps guests out of that one.
Opening a team shows its retrospectives - open ones first, the nearest first - its members, its open actions and its agreements, all on one page.
A manager renames the team, switches it between public and private, chooses when members are reminded of a retrospective, and manages its integrations. The owner can transfer the team to another member, staying on as an administrator, and a manager can delete the team with everything in it.
Anyone but the owner can leave a team. What they wrote stays, under their name, and only a manager can add them back.
A meeting nobody heard about is half empty, and one whose hour moved without a word is emptier still.
Any member starts one from the team dashboard, ahead of the meeting or at the last minute.
A template is a set of columns a retrospective starts from. The application provides five - the sprint retrospective, Start · Stop · Continue, Mad · Sad · Glad, 4L and Sailboat - read in each member's own language. The Templates tab lists them beside the team's own, which any member writes, edits and deletes.
A retrospective keeps its own copy of the columns it started with: changing or deleting a template afterwards never touches a board already started from it.
Most teams meet at the end of every sprint, and starting each retrospective by hand is easy to forget. When you start one, choose how often it repeats - every week, or every 2, 3 or 4 weeks. Finishing it then starts the next one, on the same day and at the same time in the time zone of whoever chose the repetition, with a default name and the same privacy. The lists show how often an open retrospective repeats; set it back to "Does not repeat" from Edit retrospective to stop.
Every member receives an announcement by mail, with buttons for Google Calendar and Outlook and an .ics file for everything else, and the team's chat channels post it too if it has any. A retrospective starting within the hour is not announced: the room is already forming, and a mail would only teach people to ignore the next one.
Members who have not said they will be away get a reminder shortly before the start. How long before - from 15 minutes to a day, or never - is chosen once in the team settings.
Both mails carry a link to say you will not be there, which works without signing in and asks before it acts. Somebody who is away does not hold up the count of who is ready, and is left out of reminders. The same page lets you change your mind, and anyone can mark a person present or absent from the board.
Until it is finished, any member can edit a retrospective from its menu. Moving its start mails the team the old time beside the new one, with an updated .ics that replaces the entry in most calendars. The cards already written stay where they are.
When the room is half empty, any member can send a nudge from the participants panel: a mail to everyone expected who is not marked away, saying the team is waiting. One nudge every five minutes at most, whoever sends it.
A team manager can delete a retrospective that is not finished yet, with its cards and the actions filed under it. If the team had been told about it, everyone receives a cancellation notice.
The first card written anchors everyone who reads it. On this board nobody reads anything until everybody has written.
While the board is hidden, you see your own cards and nobody else's: the others appear as blank placeholders. This is not a blur - the text of a hidden card never reaches anybody else's browser, so there is nothing to peek at. Names stay on the cards, so what the team gains is a first draft nobody was steered into, without losing who said what.
Each column of the board asks one question, the ones of the template it was started from - by default your mood, an obstacle you hit, something that went well, anything to bring up, and a thank-you. Type an idea under a column and press Enter to add it, Shift+Enter for a new line; emoji are one click away. A web address typed in a card, an action, an agreement or a comment becomes a link.
Until the retrospective is finished, you can reword a card you wrote, move it to another column - drag it, or pick the column from its arrows icon, which is how it works on a phone - or remove it. Nobody else can, and none of it leaves a mark on the board.
Press I'm ready when you have nothing more to add. The board counts how many of the people present are ready, so the room knows when to move on without anyone asking out loud. It is a signal, not a lock: you can still add a card, and take the ready back.
The participants panel shows who is here right now, who is ready and who is still writing. Mark somebody absent and they stop holding up the count; mark them present again if they turn up.
Anyone can say they will run the meeting, or let the board pick somebody at random among the people present - never the same person twice in a row. The name is shown on every screen, and that is all it does: running a retrospective grants nothing, so whoever it is can leave early and the meeting carries on.
A discussion that has no end tends to run until the meeting does. Any member can start a timer from the board, for between 1 and 60 minutes: it counts down on every screen, the projector included, and rings when it runs out. It ends nothing - the board stays as it was, and the room decides what comes next. Start it again to change its length, or stop it. A screen opened after the end shows that time is up, without ringing.
When the room is ready, any member reveals every card, for everybody at once - from the board's menu, or by starting to present. There is no personal view to reveal on your own. The board can be hidden again, which only affects cards written from then on.
Presenting shows one card at a time, large, with its column and its author - made for a shared screen.
Five reactions - thumbs up, heart, laugh, celebrate and wow - let you agree with a card without writing another one. Everyone sees the counts, never who reacted. You can put each reaction once on a card and click again to take it back. While the board is hidden you can only react to your own cards, since a counter moving elsewhere would give away that a card exists.
Cards, reactions, the ready count, the presented card, the timer: every change appears on every screen the moment it happens, without reloading anything - on a laptop, a projector or a phone alike.
A retrospective produces decisions that nobody carries. Here they belong to the team, and they come back until they are done.
An action is a piece of work the team decided to do. It is written down during the meeting, not afterwards from memory.
An action belongs to the team, not to the meeting it was decided in. Whatever is still open appears in the Actions panel of every following retrospective, under the ones decided today, until somebody completes it. Follow-through is the board's job rather than the job of whoever remembered to write it down.
A retrospective that opens on a blank board forgets what the team promised last time. When there is something to look back on, the Actions panel and the shared screen offer Review N actions: a full page, made for the projector, with every action still open - the most pressing first - and every one completed since the previous retrospective. Tick or untick them as the room goes through them, and every screen follows. It is optional, it is nobody's to run, and nothing records that it happened.
Tick an action once it is done. A completed action is a record: its priority, due date and assignees are locked, and so is deleting it, until somebody reopens it. Any member can edit, complete, reopen or delete an action; a guest only reads them.
The team dashboard lists the team's open actions across all its retrospectives. The arrow beside it opens a page that also keeps the completed ones, under their own heading, so the team can look back at what it got done. Your teams shows the open ones assigned to you, across every team you belong to, with a page listing them all.
An agreement is a rule the team decided to hold itself to - "we review pull requests before standup", "no meetings on Friday afternoon". Unlike an action it never ends, so it has no status, no assignee and no due date: it stands until somebody deletes it. Agreements are listed on the dashboard and in the Agreements panel of every board, so the team meets them again at each retrospective. Click one to reword it.
Each agreement carries its own thread of comments, the only written conversation the application holds. You can edit or delete your own comments, and a team manager can edit or delete anybody's.
The board is closed, and what was said has to stay findable - for the team, and for whoever was not in the room.
Any member finishes it, from the board's "..." menu or at the end of the presentation. The board turns read-only on every screen, every card is revealed, and nobody can add or remove anything any more. Whoever was named as running it is kept; if nobody was, it is whoever finished it.
A finished retrospective becomes its summary: when it was held and who ran it, how many ideas were written by how many people, the actions decided, every card by column with its reactions, and who was absent. It is mailed to every member, each in their own language, and can be downloaded as a Markdown file.
The summary offers to start the next retrospective straight away, so the date is set while everybody is still in the room.
Copy link, in a retrospective's menu or on its summary, gives a short address to paste in a chat. It grants nothing: whoever follows it still signs in, and still needs to be allowed in. It leads to the board while the retrospective is open, and to the summary once it is finished.
A team manager can publish a finished retrospective's summary. That creates a link anybody can read without signing in - the cards, who wrote them and the actions - which every member can then copy. Revoking the link destroys it for everyone who has it, and publishing again creates a different one. It works on a private retrospective too: publishing shares the summary, never the board.
A team manager can also mail the full summary to any address, a colleague on another team or a manager who asked. The mail is the copy: it carries no link, so there is nothing to revoke, and replies come back to whoever sent it.
Closed a retrospective too early? A team manager reopens it. The board accepts cards again, the summary and any published link are discarded, and a fresh summary is sent when it is finished again.
How you sign in, how well that is protected, and how the interface speaks to you are yours to decide.
Register with your email address and a password, and activate the account from the link mailed to you - it lasts a week, and a reminder arrives the day before it expires. Or sign up with Google or Microsoft: a Microsoft sign-up is confirmed by a mail to the address first. Somebody invited to a team gets the account with the invitation.
An account can hold several ways in at once, and the last one can never be removed. Add or remove them from your account settings.
Turn it on and every sign-in also asks for a code - from an authenticator app on your phone, or sent to your email address, or both. It applies to the account, so it is asked whether you signed in with a password, Google or Microsoft.
When you turn it on, you receive a set of single-use recovery codes, shown once. Keep them somewhere safe: if you lose your phone or your mailbox, they are the only way back in, because nobody can take two-step verification off somebody else's account.
You receive a mail when your account is signed in to from a browser it has not been used on before, and when somebody enters your password but not your second step. Active sessions lists every place you are signed in and ends any of them, or all but this one. Changing how you sign in asks you to sign in again first if your session is not recent.
Set your full name, the name you would rather be called by, your pronouns, title and photo, and change your email address - the new one is confirmed by a link sent to it. Your teammates read the name you prefer, on the board, the roster and in the mail they receive; administrators of this instance and its audit log keep your full name. Anybody who can see the members of one of your teams can open your profile from that list: your names, pronouns, title and photo, and the time it is where you are, beside how far ahead or behind them that is. Your own account overview adds the teams you belong to, how many retrospectives you joined, how many cards you posted and how many actions you hold.
The interface speaks English (United States or United Kingdom) or French, and writes dates the way that region does. Times are shown in your time zone, which your browser fills in on your first visit and you can change. The appearance is light, dark, or follows your device. Mails are written in your language and your time zone too.
Notifications, in your account settings, lets you turn off three kinds of mail a retrospective sends by itself: the announcement that one is planned, the reminder before it starts, and the summary once it is finished. Each switch is saved as soon as you flip it. You are still told when a retrospective you heard about moves or is cancelled, and mail about your account's security, your invitations, your role in a team, or a nudge a teammate sends by hand always arrives.
Download your data, in your account settings, prepares a copy of your account and of everything you wrote here - cards, actions, agreements, comments - with just enough around each piece to read it, and never what somebody else wrote. It takes a moment: a link arrives by mail, valid for seven days. Asking for it requires a recent sign-in.
When something you can see changes, a mark appears on your avatar and a short notice on your list of teams and your team's dashboard, never during a meeting. What's new, in the avatar menu, lists every change, newest first; opening it or dismissing the notice puts both away until the next change.
You can delete your account from its settings once you own no team - transfer or delete those first. Your name, address, photo and ways to sign in are removed. What you wrote stays, because it is part of a conversation your teammates took part in, but it is shown as written by a removed account.
Some teams live in their chat and their scripts as much as in a browser. The application posts to the chat, and the API lets those tools act for you, never beyond what you could do yourself.
A team manager adds chat channels from the team settings: a Slack incoming webhook or a Teams workflow address, a test message posted before it is saved, and a choice of what it carries - the announcement, a change of time, the reminder, a nudge, the summary. Posts go out beside the mails, never instead of them, in the channel's own language and time zone. A post never repeats what was written on the board and never names a private retrospective: it says what happened and links to it, so everything else stays behind a sign-in. Every member can see the channels.
A chat bot, an agent, a script grown into a service: a team manager registers it from the team settings, under Integrations, and receives the identifiers it signs in with - its secret shown once. The integration belongs to the team, so a colleague leaving costs it nothing.
Registering an integration authorises it over nothing. The first time it acts for you, you see a screen naming exactly what it asks for, and you allow or deny it. It is asked again whenever it wants more. You can withdraw it at any time from Authorised integrations in your account settings, and it loses you anyway when you leave the team.
What an integration or a token may do is cut into scopes - reading or writing the team, its retrospectives, its actions, its agreements. A scope is a ceiling, never a key: a token that may read retrospectives reads exactly the ones you could open in your browser, and nothing else.
From API tokens in your account settings, create a token with a name, its scopes and, if you like, an expiry. It is shown once and kept only as a hash, so copy it then; if you lose it, revoke it and make another. Creating one asks you to sign in again if your session is not recent. A token lets a script call the API as you, and it is not a way to sign in.
Claude, ChatGPT or Cursor can read your team's retrospectives, actions and agreements and answer your questions about them, through the MCP server at https://retro.jdl.li/mcp. It reads and never writes, and it sees exactly what you see. For an assistant in your browser, a team manager first registers it as an integration with the redirect address the assistant gives; you then paste the integration's identifier into the assistant's connector settings and allow it like any other integration. For an assistant on your computer, such as Claude Code, a personal token with the read scopes is enough.
Every endpoint, its parameters and its answers are described in a public reference, readable without an account. Open the API reference
An account is free, and takes an email address or a Google or Microsoft account.