card-conventions
Skill130 starsChanged 8 days agoMITView on GitHub
Card conventions for Platypus frontend lists. Use when adding or changing a list, card or ⋮ menu in apps/frontend.
In this repository38 other files it gives its agents.
- .agents/skills/humanizer/AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- add-api-endpointSkill
- api-conventionsSkill
- ask-mattSkill
- codebase-designSkill
- code-reviewSkill
- database-schema-changesSkill
- diagnosing-bugsSkill
- docs-auditSkill
- domain-modelingSkill
- form-conventionsSkill
- grillingSkill
- grill-meSkill
- grill-with-docsSkill
- handoffSkill
- humanizerSkill
- implementSkill
- implement-specSkill
- improve-codebase-architectureSkill
- platypus-toolsSkill
- pre-release-checkSkill
- prototypeSkill
- prSkill
- researchSkill
- retroSkill
- setup-matt-pocock-skillsSkill
- tddSkill
- teachSkill
- to-questionnaireSkill
- to-specSkill
- to-ticketsSkill
- triageSkill
- ui-conventionsSkill
- wait-whatSkill
- wayfinderSkill
- wizardSkill
- writing-for-agentsSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 2 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 70
Leave out a prop the User lacks the permission for.
- Very specific: line 94
Every action lives in the ⋮ menu, so the card itself carries only content and badges.
- Very specific: line 100
Every card in a list is one height, whatever it holds, so a paged list keeps its height.
- Distinctive: line 129
After the create, go to the clone's edit page.
- Very specific: line 130
Use CloneDialog (components/clone-dialog.tsx) for the dialog.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Frontend Card Conventions
The Workspace home's Agents, Boards, Triggers and Skills cards are the reference implementation; a new card list reads the same way.
Build every card list on EntityCardList (components/entity-card-list.tsx).
It owns the grid, the loading/error/empty ladder, and the one-height card, so a
list fills in the slots instead of re-building them: media leads the card,
badges fills the title row, meta adds the lines under the description,
actions puts a primary button before the menu, menu is the RowActionsMenu,
href is the card's click, and isShared adds the Organization badge and
locks the click. Use its groups for a Pinned section. A list on a settings
screen (beside the settings menu) sets singleColumn; the two-column grid is
for full-width screens.
What clicking a card opens
A card opens the item's main view, unless it carries an explicit primary button. The primary button is then the card's one action, and the card itself stays a plain, non-clickable surface.
| Item | Card click opens | Why |
| Agent | nothing — New chat is primary | the visible button is the main action |
| Board | the board | its main view |
| Trigger | its runs, filtered to it | the runs are what a Trigger produces |
| Skill | Edit | it has no view beyond its settings |
An item with no view beyond its settings opens Edit. Edit always lives in the ⋮ menu too, so it is reachable from every card.
The ⋮ menu order
Edit
item-specific (Pin, Promote to organization, Manage attachments, …)
Clone
Enable / Disable
───────── separator
DeleteSkip the items that do not apply and keep the rest in this order. Delete takes the destructive style.
Build every list card's (and run row's) ⋮ menu with RowActionsMenu
(components/row-actions-menu.tsx), never a hand-built DropdownMenu; it
enforces the order and the separator.
<RowActionsMenu
name={item.name} // "Actions for <name>"
shared={isShared} // Edit in org settings + Detach
edit={{ href }}
actions={[ // RowAction: href (a link) or onSelect
{ label: "Pin", icon: <Pin />, onSelect, disabled },
{ label: "View runs", icon: <List />, href },
]}
clone={{ onSelect }}
toggle={{ enabled, onSelect, disabled }}
remove={{ onSelect }} // Delete, or Detach when `shared`
/>Leave out a prop the User lacks the permission for; the component drops the ⋮ when nothing is left, so a list needs no "has menu" check of its own.
Shared items
A Shared resource attached to a Workspace (scope: "organization") is locked
there. Its card gets a ⋮ menu with:
- Edit in org settings, linking to the item in the organization settings,
only for a User who can manage Shared resources (
canManageSharedResource); - the item-specific items that still apply (Pin);
- Clone, for a Shared Agent, only for a User who can create one here
(
canCreateAgent); - separator, then Detach, in Delete's place.
Pass shared to RowActionsMenu for both labels.
Detach is always a named menu item that opens the shared Detach dialog
(DetachSharedDialog). A Shared card's own click opens its main view if it has
one, and otherwise does nothing. When the User has no items to offer, the card
has no ⋮ button at all.
Icons and badges
Every action lives in the ⋮ menu, so the card itself carries only content and
badges. A disabled item shows a Disabled badge in its title row — the secondary
Badge with the Pause icon and the word Disabled, as Trigger cards draw it.
Card height
Every card in a list is one height, whatever it holds, so a paged list keeps
its height. Reserve line boxes rather than tuning a min-h on the card:
- the title row is
min-h-badge, with or without badges, and the name is one line:truncated inside amin-w-0ItemTitle, with atitle; - the description always renders, clamped and reserving the same lines
(
line-clamp-2 min-h-[2lh];line-clamp-1 min-h-[1lh]on a card with nothing after it, such as a Board's), withtitleholding the full text; - each line after it is one line tall even when empty (
h-4), its texttruncated with atitle; a list of names becomes a count with a tooltip (CountTooltip).
The list's skeleton reserves the same boxes, so it lands without a jump.
Clone
Clone (Agents and Triggers) runs in the browser: send the ordinary create request
with only the fields the create takes, never the item's identity (id,
timestamps, scope) or its run and token state.
- The clone is named
<name> (copy), editable in the Clone dialog. - Tokens and other issued credentials stay with the original; a clone that needs one gets its own.
- A cloned Trigger starts disabled, so it never fires beside its original.
- A cloned Inbound Trigger has no token (
issueToken: falseon the create) until the Owner issues one. - An Agent clone keeps only the Sub-Agents, Skills and Tool sets visible in the
Workspace (the lists the cards count), and the Clone dialog's
notesays how many of each it leaves out. - After the create, go to the clone's edit page.
- Use
CloneDialog(components/clone-dialog.tsx) for the dialog.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

