Boards
Boards hold lists and lists hold cards. Any card can have subcards of its own, so a board can be a plan, a backlog, a sprint or a roadmap.
Boards, lists and cards
A board is a set of lists (columns), and each list holds cards that you drag
between and within lists. Every board also has a short code, such
as qvm-v1-rem, that names it in links, the API and the CLI.
A card can carry:
- a Markdown description, a checklist and comments
- a priority from low to critical, a start date and a due date
- flags: flagged, blocked, needs review, waiting, starred, stuck
- tags, one or more assignees, a cover colour, web links and attachments
- a timer, with time spent against an estimate
A flag says who clears it. Filled, with a solid icon (and a raised hand on the open card), means nothing on the board will clear it: a person has to act. Outline only, and clears itself on the open card, means something already in motion will: a blocked card whose blocker is still open is unblocked when that card is done, and a card awaiting verification on a board with a verify pool is decided by a verifier agent. Waiting, flagged and stuck always need a person; a starred card asks nothing of anyone. Hover a flag to see who clears it.
@name in a description or comment mentions someone who can see the
board, and they get an email (and a desktop notification, if they have those on). Lists can have a colour, a WIP limit, and their own
sort order or grouping.
An open board keeps up by itself: a change anyone makes shows at once. As a check, the page also asks the server whether it has missed anything when you come back to its tab, and once a minute while you are looking at it, and draws the board again if it has. If the server stops answering while the page thinks it is connected, it says This board may be out of date, tries to reconnect, and offers Reload.
Subcards and epics
Any card can have subcards. It becomes a board of its own, with lists from a template, and those subcards can have subcards too, to any depth. A common pattern is for top-level cards to be epics and their subcards to be the actual tasks.
Each card summarises everything beneath it: how many tasks are done, the earliest start and latest due date, whether anything below is blocked or overdue, and a health state. My work lists everything assigned to you across every board and level, after what is waiting on you: the overseer's alerts first, when a board has it on (Automations, Overseer). Open cards with no due date aren't listed there; its header says how many there are, and completed ones are under Show completed. The API, the CLI, the MCP tools and the AI chat still see every assigned card.
Cards can also depend on each other ("blocked by" and "blocks"). A card waiting on unfinished work shows a blocked badge.
Waiting on you
My work opens with everything that needs you, oldest first: questions waiting on you, cards flagged for your review, failed runner jobs, cards in progress with nobody on them and Scheduled cards not started by the end of their day, on the boards you own. The header counts them on every page.
The boards show them too. An epic's card carries ⏸ 2 waiting on you
for what in its subcards is on you, and a line naming each with its reason, such as Waiting on you: #2063 (a question), #2075 (3 rejections). Questions waiting on
other people show in the line only (and 1 on someone else), never in the count. Click the
chip to open the sub-board showing only those cards. Open the epic and its subcard list
marks the same rows: each one waiting on you carries a badge with its reason (to verify, a question),
one waiting on someone else a quieter mark naming who it waits on, and the line's #ids above the list jump to their rows.
A card further down marks the subcard it sits under, with its own #id. The board's header says
Waiting on you: N
for the whole board and opens My work kept to it. slipdock card
and slipdock board
print the same line.
A question addressed to you also raises an alert in the header that only you see, and an email, and so does a card of yours past its board's rejection limit. Turn either off in account settings, under When something waits on you.
Some of them you won't act on. The × on a row dismisses it from your own list. It stays off for that reason only: a second failed job, a new question or the flag put on again brings the card back. To keep a whole board quiet, choose Mute on my Waiting on you in the board's … menu. Nothing from that board or any board under it shows on your list, and a strip on the board says it is muted. Muting is not archiving: the board works as before, for you and everyone else.
Both are yours alone. Nobody else's list changes. The list always says what it is
hiding (3 waiting · 2 dismissed · 1 board muted), and Show
lists them with Undo
and Unmute. From the CLI it is slipdock work --dismiss, --dismissed, --undismiss,
--mute
and --unmute.
Desktop notifications
Slipdock can show a notification on your computer when a card starts waiting on you,
a question is routed to you, or somebody @mentions you: the same things
that email you, which they still do. Turn on
Desktop notifications on this browser
under Account → Preferences;
the browser asks whether to allow them. It is off until you do.
They come through a Slipdock tab you have open, in the background too, and with several tabs open you get one, not one each. With no tab open nothing shows, on purpose: Slipdock doesn't reach you when you have closed it. Clicking one brings the tab forward and opens the card. You get none for what you did yourself, nor while you are looking at that card.
The setting is kept in this browser, because the browser's permission is too: turn it on in each one you use. If notifications are blocked for the site, or the browser can't show them, the toggle says so; unblock them in the browser's site settings. Only the web app shows them; the API, the CLI and agents have nothing to set.
Views
The same cards, shown different ways. Switch between them from the view menu, or press v:
- Board: lists side by side.
- Swimlanes: a grid with any attribute on each axis, such as assignee by priority.
- Table: a sortable row per card, with editing in place (custom fields too) and CSV export. Whoever may change the board's fields has a + at the end of the header to add one as a column.
- Timeline: a Gantt chart by start and due date, with dependency lines.
- Calendar: a month or week, by due date.
- Outline: the board as a tree of cards and subcards.
- Narrative: what happened to the cards over a date range, told as prose.
- Prioritise: a ranked table where priority, votes and scores are edited in place.
The filter bar narrows any view by text, tag, priority, flag, due date, origin (who or what made the card: a person, the sweep, an agent's job…) and the board's custom fields, each by its type: options for a choice field, a minimum rating, a min and max for a number or formula, a date range, or words a text field contains. A card with no value for a field is hidden by its filter. A view with its filters can be saved, shared, or published as a public link.
On the board, a search or filter shows its matches as one list down the page, grouped by the list each card is in, in board order, so none sit off the side of the screen. Each heading gives the list's name and how many match; click it to fold or unfold the group. Done, and any list with more than 10 matches, starts folded, and the lists with no matches are named on one line at the end. Click a row to open the card, or use its arrow to move it to another list. Board in the banner shows the matches in their columns instead, for that search; Clear brings the whole board back.
Card views
Choose what you see of a card: each part of its front, each section of the opened card and each property and group of its sidebar, assignees and dates included. Card views are yours alone and kept on the server, so they follow you to every browser. Nobody else's cards change.
- Levels: global, then a board, then one list on it. Each states only what it changes, Show or Hide, and inherits the rest from the level below.
- Named card views: save a set under a name and apply it at any level, with that level's own choices on top. A saved view can name one too, under Display → Card view for this view.
- Show all: a switch in the Display menu that shows every part of the card front on this board while it is on.
- The handoff note: show Handoff note (2 lines) or Handoff note in full to read where a card was left without opening it.
- Fields: the custom fields switched to Show on cards, as one element, so a card view can hide them all.
Open them from the Display menu's Card views…, from Card view… in a list's ⋯ menu, or at Card views. The board's own settings always win: a card view never brings back what a simple board or a list hides. The list level applies on the board and swimlanes; the timeline, calendar and outline use the board's.
Planning
- Sprints: a sprint board where every card is a sprint, with burndown and velocity charts.
- Roadmaps: give lists a category (to do, in progress, done, dropped, triage, scheduled, housekeeping) and a horizon such as Q1 2027; add milestones; colour any view by list, priority, health, assignee or tag.
-
Recurring cards: Repeat…
on a card (or
every: monin quick add) makes it repeat: daily, weekdays, weekly, monthly on a date or the last Friday, yearly, or some days after the last one is done. The next one is a copy of the current one, made in your timezone, and one still open has its due date moved on rather than another piling up. On a board with a Scheduled list (a list in the Scheduled category; the Slipdock and Simple templates have one, after To Do) the next one is made there unless you chose a list: recurring agent work waits there for the runners' rules to send it on its day, out of your To Do, and nobody picks from it. One still there after its day is on your Waiting on you. Duplicate makes the same copy by hand. -
Snooze: Snooze…
on a card, or z
over it on the board (z z
is tomorrow), hides a card that isn't actionable yet until a day, then brings it back
at the top of its list. Each list's foot counts what is snoozed in it and shows them,
faded, on a click.
snooze: monin quick add does it as you add the card. - Tidy up… in the board's menu looks the board, and every board beneath it, over for rot: epics whose work is all done but still open, cards in Done never ticked off, cards stuck in progress, epics with no next step, long-overdue cards, likely duplicates, wiki pages nothing links to and links with a typo. Each comes with the evidence and a proposed fix to apply or dismiss; a dismissal is remembered, and an applied fix can be undone. Proposed handbook edits on one section say which others share it, and apply one after the other, each as its own lines. Board settings → Gardening picks the checks and how long before something counts.
- Findings: what agents' passes learned on a board, and what went wrong (a rejection, a job that timed out). Triage and the server handle them; what waits on you is a rule drafted from one that recurs. Its tile on the board's Agentic Building page counts those proposals and opens the board's Findings page (a card's Findings link opens it too), which leads with them, each linking the Tidy-up page to apply or dismiss. Below, for reference, the findings: filter by status, kind, source, card or date, and promote a finding into the handbook, merge repeats or dismiss it. A finding a waiting proposal names is waiting on a person, out of the open list (its own filter), since deciding it would close that proposal. A card with open findings says how many, and links there.
- Ledger: what shipped on a board since 07:00 yesterday, a line per card completed, accepted, merged or deployed, with who accepted it, its evidence, the should rules it missed, where its time went and what its jobs spent, grouped by day. Its tile on the board's Agentic Building page opens it; Look closer opens a card with its chat. Ten minutes' reading instead of approving each card.
-
Batched releases: Agentic Building → Releases
on the root board chooses how merged work reaches production: with each merge, or in
batches, with an age cap in hours for how long the oldest merge may wait. A merged
card is still done, and Release now
is how a person ships it; on a batched board it shows merged, awaiting release
until a deploy ships it, then released
with the time. A release runs when a card of priority critical
merges, when the oldest merge reaches the age cap, or when the board's owner presses Release now, on the board's release banner or on that page. The banner's
merged, awaiting release
count opens the list of the cards the next release takes; once it starts,
A release is running
opens the list of the cards it is shipping. Releases run one at a
time: a runner runs the board's release command, and anything asked for meanwhile
becomes one follow-up. While one runs, the banner's follow the build
opens the run it started. A runner runs the default
gh workflow runcommand itself; any other runs only if that machine'slocal.jsonlists it, word for word, underrelease_commands. A red release (its suite, build or deploy failing) holds nothing: the banner says the last release failed, with the failing run, and a critical fix card goes on top of the ready list. Release now and the age cap still release, with a warning that names the failing run. The fix card's merge clears the failure and releases; the fix card accepted with no code (a cause outside the repository) clears it and releases nothing: the batch then goes at the next trigger. A deploy that succeeds at the failed sha or past it (the failed run rerun, or a later release) clears it too, and archives the fix card if nothing had started on it. - Who gets asked: Agentic Building → Who gets asked names a person for each kind of question an agent asks (product, UX, security, release, anything else). Their waits go to that person, and to the board's owner when nobody is named. Beside it, An approval's proposal stands after (off until set; 4h, say) lets an unanswered wait of kind other that offers choices and proposes one go ahead with its proposal: the card says so and the person asked is told. Product, UX, security and release questions, and a wait that proposes nothing, always wait for a person. On a board whose Sensitivity is high, no wait goes ahead by its proposal.
- Sensitivity: Agentic Building → Sensitivity on the root board says how carefully agents proceed on the whole tree: low, medium (the default) or high. A sub-board follows its root and can't set its own. At low, agents make the choices they would otherwise ask you about (between implementation options, say), and no card is held for a person by its profile. At medium, nothing changes. At high, agents ask before any material choice, feature cards are held for you before merging as schema and security ones are, and an approval never goes ahead by its proposal. At every level a security question is still asked. A joke-telling internal tool might be low; a system holding a customer's personal data, high.
- Profiles: one word for what kind of change a card is (security, schema, feature, fix, docs or spike), off until Agentic Building → Profiles turns them on. An agent starting a card with none gets one from its paths and tags, strictest first; the card's front shows it as a chip, and the card panel's Profile select sets or clears it. When an agent completes the card, the profile is checked again against what the change touched, and raised, never lowered: a migration or a security path raises even one you set, and more than 400 changed lines makes an agent's or an inferred fix or docs a feature (a fix you set keeps it). Human tier, by profile, on the same page, says whether an agent's acceptance merges the card, merges it onto the ledger or holds it for a person: by default schema and security changes wait for you. The tree's Sensitivity overrides it: low holds nothing, high holds features too. A spike merges nothing: its prototype stays on its branch.
- Incidents: when a card broke something that shipped, its Incident line, under Dependencies, names the card that shipped it and the check that should have caught it (bar, rules, CI, verify, audit or none). The causing card then shows the incident it caused, and carries a finding naming that check.
- Status updates: on track, at risk or off track, with a note, shown beside the health the board works out for itself. On a card agents work, the board posts each step itself (Writing code, Verifying, Merged, Shipped to production…), quietly, at the top of the card; a finished card shows there how it ended (done when, accepted by whom, what merged and shipped) instead of a next step. With delivery metrics on, a merged card that no succeeded deploy names after 20 minutes (the board's Not shipped after setting) shows an amber Merged 1a634c6 · not shipped after 45m and raises one warning alert on the board, whether or not the overseer is on; the deploy that names it clears both. On a board with batched releases the minutes count from the card's release, not its merge, so a card waiting for its batch raises nothing. How long ago each one was ("18 min ago") keeps up with the clock while the card is open.
- Custom fields (numbers, ratings, choices, dates, text, links, checkboxes, people) and formula fields with RICE and ICE presets. Switch a field's Show on cards in board settings to put its value on the card front as a chip; a compact card shows the first two. A choice field's Allow several lets a card hold several of its options: it is in each of their swimlanes, its filter keeps a card holding any of those picked, and a formula adds their weights. A person field holds several of the board's members: filter it by Me or anyone on the board, or make it a swimlane axis. Someone who leaves the board stays on their cards, marked as a former member, and can't be picked again.
- Budget voting: everyone spends a set number of votes across the board.
- Typed links between cards on any board, so a goal card can collect the work contributing to it.
Getting around fast
- Quick add: press q or use the header box and type a line such as "call the printers about the banners friday, urgent". It becomes a card on your default board. After a semicolon comes the card's description, after two a comment: "Fix the login; seen on Safari only;; ask Sam first".
- Ctrl-P opens the command palette, and Ctrl-O opens any card by its title or number. Press ? for every keyboard shortcut.
- Favourites: heart a board, list, card or saved view to keep it two taps away.
- Select in the board's toolbar picks out cards to change all at once: click a card to pick it, Shift-click another to pick every card between the two (down a list, then across), or press the tick in a list's header to pick every card it shows (again to let them go). Select all in a list's ⋯ menu turns select mode on with every card the list shows picked; in select mode it adds them to what is already picked, and never lets any go.
- On a phone, the board becomes one list at a time, with navigation and quick add in thumb reach.
Templates, moving and archiving
- Templates are named sets of lists for new boards and sub-boards.
- A card moves to another board with its subcards, from the same menu that moves it between lists.
- Archiving a card or a whole board puts it away without deleting anything, and it can be restored.
- Boards can be imported from Trello or another Slipdock server, and everything you have can be exported, under Account → Import & export.