refactor(dashboards): keep the authoring reference in Mothership's skill

Mothership's create-dashboard skill now carries the dashboard syntax like
every other worker skill, so the builtin-create-dashboard workspace skill,
its source, and its flag gating are removed.

Author: TheodoreSpeaks

apps/sim/app/o/[organizationId]/home/components/composer/composer.tsx

@@ -25,7 +25,6 @@ import {
 } from '@/app/workspace/[workspaceId]/home/components/user-input/components/prompt-editor'
 import { organizationSkillOptions } from '@/app/workspace/[workspaceId]/home/components/user-input/components/skills-menu-dropdown/organization-skill-options'
 import type { ChatRequestMode } from '@/app/workspace/[workspaceId]/home/types'
-import { useFeatureFlag } from '@/app/workspace/[workspaceId]/providers/feature-flags-provider'
 import type { useFileAttachments } from '@/app/workspace/[workspaceId]/w/[workflowId]/components/panel/components/copilot/components/user-input/hooks/use-file-attachments'
 import { SKILL_CHIP_TRIGGER } from '@/app/workspace/[workspaceId]/w/[workflowId]/components/panel/components/copilot/components/user-input/utils'
 import { getSkillsQueryOptions } from '@/hooks/queries/skills'
@@ -76,7 +75,6 @@ export function Composer({
   attachedFilesRef.current = files.attachedFiles
   const imagesOnly = requestMode === 'assistant'
   const { organization } = useOrganizationContext()
-  const dashboardsEnabled = useFeatureFlag('dashboards')
   const { data: allWorkspaces = [] } = useWorkspacesQuery(!imagesOnly)
   const workspaces = (imagesOnly ? [] : allWorkspaces).filter(
     (workspace) => workspace.organizationId === organization.id
@@ -90,8 +88,7 @@ export function Composer({
         workspaces.map((workspace, index) => ({
           ...workspace,
           skills: skillQueries[index].isPlaceholderData ? [] : (skillQueries[index].data ?? []),
-        })),
-        dashboardsEnabled ? [] : ['builtin-create-dashboard']
+        }))
       )
   const editor = usePromptEditor({
     workspaceId: '',

apps/sim/lib/dashboards/README.md

@@ -1,10 +1,10 @@
 # Table-backed dashboards
 
-Each workspace has at most one dashboard, a separate resource with its own sidebar page, resource tab, and Mothership `dashboards get` / `dashboards set` commands. Sim builds it: the page shows an empty state until the first save, and there is no create, rename, move, or folder operation. A partial unique index (migration 0392) keeps one live dashboard per workspace; lifting the limit means dropping it. Storage reuses workspace files with MIME type `text/x-sim-dashboard`; the backing `.dashboard` suffix is hidden from display names. The built-in **create-dashboard** skill documents the syntax without example dashboards or prescribed layouts. Sharing is deferred.
+Each workspace has at most one dashboard, a separate resource with its own sidebar page, resource tab, and Mothership `dashboards get` / `dashboards set` commands. Sim builds it: the page shows an empty state until the first save, and there is no create, rename, move, or folder operation. Dashboards live in the `dashboard` table (migration 0392) with their own id; a unique index on `workspace_id` keeps one per workspace, and dropping it allows several. Saves compare the stored `revision`, so a concurrent edit is never overwritten. Mothership's **create-dashboard** skill documents the syntax. Sharing and fork copies are deferred.
 
 The implementation has three boundaries:
 
-- `spec.ts` validates a bounded YAML document and normalizes ECharts options through the existing `.chart` safety rules. It rejects unknown layout/source keys and reports errors in the viewer and the file VFS `compiled-check` path.
+- `spec.ts` validates a bounded YAML document and normalizes ECharts options through the existing `.chart` safety rules. It rejects unknown layout/source keys; saves reject invalid YAML and the viewer reports panel errors.
 - `table/analytics` computes exact aggregates over authorized table rows. The internal POST `/api/table/[tableId]/analytics` is a session-authenticated adapter for `tables.rows.analytics`, requiring the current viewer's workspace read role and `tables.use`. The operation is session-only because this release's sole query caller is the workspace renderer. Public/versioned query APIs, workflow/executor callers and log queries are deferred. Dashboard APIs and Mothership tools share the dashboard application operations.
 - `components/dashboards` owns EMCN layout, controls and states. `components/charts/echarts-view.tsx` also renders existing `.chart` files, using the local EMCN tokens for its canvas theme. `.chart` retains its existing sampled source behavior; dashboard aggregation is performed on the server.
 
@@ -45,7 +45,7 @@ Queries reuse the table predicate compiler and the existing read-only repeatable
 
 Bounds: 128 KB source, 48 blocks, 4 layout levels, 2 grouping fields, 8 measures, 12 projected columns, 500 result rows and 8 KB per returned row. Limits apply after aggregation. An explicit limit yields a labeled top-N result; unrequested group overflow is an error. API rate admission is per viewer. Errors are never turned into successful zeros. Only visible tabs mount their query observers; identical queries share React Query cache entries for one minute, and Refresh requests fresh data.
 
-Public file shares show a workspace-only message and issue no table requests. This release does not embed live dashboards in public HTML pages or export them as self-contained HTML. Row/query data is not persisted inside the dashboard file.
+Dashboards are workspace-only: there is no public sharing, embedding in public HTML pages, or self-contained export. Row/query data is never persisted with the dashboard.
 
 ## Validation
 
@@ -67,21 +67,12 @@ global switch. Local development uses `DASHBOARDS=true` in the app's ignored
 environment file.
 
 The server resolves the flag for the canonical workspace organization. It gates
-dashboard operations, table analytics, UI entry points, and the
-built-in authoring skill. Mothership receives that availability per turn and
-persists it for continuation; disabled runs omit dashboard commands and skills.
+dashboard operations, table analytics, and UI entry points. Mothership receives it per
+turn as the `dashboards` entitlement and persists it for continuation; runs without it
+omit the dashboard commands, the create-dashboard skill, and dashboard prompt text.
 The Sim server checks current availability on every dashboard operation, including
 calls from a run admitted before the flag changed.
 
 Apply both repositories' additive migrations and deploy the companion worker
 before enabling the flag. Sharing and a tool for capturing the user's displayed
 data are deferred.
-
-
-## Storage
-
-The workspace dashboard is a workspace file at the root with content type
-`text/x-sim-dashboard`. Files listings, file search, and workflow file pickers exclude that
-content type, and only the dashboard use case may write it, so the backing file never appears
-or changes as an ordinary file. Versions, billing, cleanup, and complete workspace copies
-still include it.

apps/sim/lib/dashboards/authoring.ts

@@ -14,13 +14,11 @@ import { beforeEach, describe, expect, it, vi } from 'vitest'
 
 const { hoisted } = vi.hoisted(() => ({
   hoisted: {
-    flag: vi.fn(),
     getById: vi.fn(),
     update: vi.fn(),
   },
 }))
 
-vi.mock('@/lib/dashboards/feature-flag', () => ({ requireDashboardsEnabled: hoisted.flag }))
 vi.mock('@/lib/uploads/contexts/workspace', () => workspaceUploadsMock)
 vi.mock('@sim/platform-authz/workspace', () => workspaceAuthzMock)
 vi.mock('@sim/audit', () => auditMock)
@@ -63,24 +61,12 @@ const skill = {
 
 describe('skill application use cases', () => {
   beforeEach(() => {
-    mocks.flag.mockResolvedValue(undefined)
     mocks.loadContext.mockResolvedValue(workspace)
     mocks.resolvePermission.mockResolvedValue('read')
     mocks.getById.mockResolvedValue(skill)
     mocks.update.mockResolvedValue({ ...skill, content: '# Updated' })
   })
 
-  it('refuses the dashboard authoring skill when dashboards are disabled', async () => {
-    mocks.getById.mockResolvedValue({ ...skill, id: 'builtin-create-dashboard' })
-    mocks.flag.mockRejectedValue(new Error('Dashboards are not enabled'))
-    await expect(
-      getSkillUseCase.execute({
-        principal: { kind: 'session', userId: 'reader' },
-        input: { workspaceId: workspace.workspaceId, skillId: 'builtin-create-dashboard' },
-      })
-    ).rejects.toThrow('Dashboards are not enabled')
-  })
-
   it.each(['skill-1', 'builtin-research'])(
     'reads %s with read permission in the asserted workspace',
     async (skillId) => {

apps/sim/lib/skills/application/use-cases.test.ts

@@ -11,7 +11,6 @@ import {
   ForbiddenOperationError,
 } from '@/lib/core/application'
 import { OrchestrationError } from '@/lib/core/orchestration/types'
-import { isDashboardsEnabled, requireDashboardsEnabled } from '@/lib/dashboards/feature-flag'
 import { getSkillActorContext, listSkillEditors, type SkillEditor } from '@/lib/skills/access'
 import { skillDelegationPolicy } from '@/lib/skills/application/authorization'
 import { skillOperations } from '@/lib/skills/application/operations'
@@ -212,9 +211,6 @@ export const listSkillsUseCase = defineAuthorizedWorkspaceUseCase({
   async execute({ input, context }) {
     const page = await listSkillSummariesPage({
       workspaceId: context.workspaceId,
-      excludedBuiltinIds: (await isDashboardsEnabled(context.workspaceOrganizationId))
-        ? []
-        : ['builtin-create-dashboard'],
       search: input.search,
       sortBy: input.sortBy,
       sortOrder: input.sortOrder,
@@ -243,9 +239,6 @@ export const listAvailableSkillsUseCase = defineAuthorizedWorkspaceUseCase({
     const skills = await listSkillsForUser({
       workspaceId: context.workspaceId,
       userId: requirePrincipalSubjectUserId(principal),
-      excludedBuiltinIds: (await isDashboardsEnabled(context.workspaceOrganizationId))
-        ? []
-        : ['builtin-create-dashboard'],
     })
     return { skills }
   },
@@ -261,11 +254,6 @@ export const getSkillUseCase = defineAuthorizedWorkspaceUseCase({
   resolveContext: ({ input }: { input: GetSkillInput }) =>
     resolveSkillContext(input.workspaceId, input.skillId),
   authorizationOptions,
-  async authorizeResource({ context }) {
-    if (context.skill.id === 'builtin-create-dashboard') {
-      await requireDashboardsEnabled(context.workspaceOrganizationId)
-    }
-  },
   async execute({ context }) {
     return { skill: context.skill }
   },

apps/sim/lib/skills/application/use-cases.ts

@@ -1,5 +1,3 @@
-import { DASHBOARD_AUTHORING_CONTENT } from '@/lib/dashboards/authoring'
-
 /**
  * Built-in (template) skills that ship with every workspace.
  *
@@ -192,13 +190,6 @@ You can attach an existing workflow to a table so it runs once for every row, wi
 `
 
 export const BUILTIN_SKILLS: readonly BuiltinSkill[] = [
-  {
-    id: 'builtin-create-dashboard',
-    name: 'create-dashboard',
-    description:
-      'Build a live table-backed dashboard with YAML, ECharts, time filters, rows, and tabs.',
-    content: DASHBOARD_AUTHORING_CONTENT,
-  },
   {
     id: 'builtin-connect-integration',
     name: 'connect-integration',