@databrainhq/mcp-server 0.2.3 → 0.2.4

package/README.md

@@ -197,12 +197,7 @@ Prompts are guided workflows your AI assistant can follow. Ask for them by name
 |--------|-------------|
 | `embed-existing-dashboard` | Embed a dashboard you've already built in the Databrain UI |
 | `embed-blank-dashboard` | Multi-tenant setup — each client gets their own dashboard |
-| `enable-metric-creation` | Let end-users build metrics (drag-and-drop or AI chat) |
 | `brand-my-embed` | Customize colors, fonts, charts, card styling, responsive layout |
-| `localize-embed` | Multi-language support, translations, calendar format |
-| `secure-my-embed` | Permissions, access controls, role-based guest tokens |
-| `setup-filters` | Dashboard filters — dropdowns, date ranges, server-side filters |
-| `manage-embeds` | List, compare, and review existing embed configurations |
 | `query-data` | Query your data with natural language, manage the semantic layer |
 | `populate-semantic-layer` | Auto-generate table/column descriptions to enable natural language querying |
 
@@ -210,7 +205,7 @@ Prompts are guided workflows your AI assistant can follow. Ask for them by name
 
 ## Tools Reference
 
-24 tools organized by category. Your AI assistant picks the right tools automatically — you don't need to call them directly.
+21 tools organized by category. Your AI assistant picks the right tools automatically — you don't need to call them directly.
 
 > **Note:** Infrastructure creation (datasources, datamarts, workspaces) is done in the **Databrain UI**. However, the **semantic layer** is populated via MCP tools (`get_semantic_layer`, `update_semantic_layer`) — the AI auto-generates descriptions from column names.
 
@@ -239,17 +234,15 @@ Prompts are guided workflows your AI assistant can follow. Ask for them by name
 </details>
 
 <details>
-<summary><b>Embeds</b> (7) — Create, configure, and customize embedded dashboards</summary>
+<summary><b>Embeds</b> (5) — Create, configure, and customize embedded dashboards</summary>
 
 | Tool | Description |
 |------|-------------|
 | `create_embed` | Create an embed (standard or multi-tenant with `clientId`) |
 | `list_embeds` | List embed configurations |
 | `get_embed_details` | Get full embed details by ID |
-| `update_embed` | Update embed config (raw patch) |
+| `update_embed` | Update embed config — access settings, options, filters, i18n (via `body` passthrough) |
 | `customize_embed_theme` | Presets, colors, fonts, cards, chart appearance, responsive layout |
-| `configure_embed` | UI options, filters (string/date/number/SQL), i18n, translations |
-| `diagnose_embed_health` | Diagnose malformed embeds + validate runtime readiness |
 </details>
 
 <details>
@@ -258,7 +251,7 @@ Prompts are guided workflows your AI assistant can follow. Ask for them by name
 | Tool | Description |
 |------|-------------|
 | `generate_guest_token` | Generate a frontend guest token (server-side) |
-| `generate_embed_code` | Generate integration code (React, Next.js, Vue, Angular, Svelte, vanilla JS) |
+| `generate_embed_code` | Generate integration code (React, Next.js, Vue, Angular, Svelte, SolidJS, vanilla JS) |
 </details>
 
 <details>
@@ -272,12 +265,11 @@ Prompts are guided workflows your AI assistant can follow. Ask for them by name
 </details>
 
 <details>
-<summary><b>Discovery & Orchestration</b> (2)</summary>
+<summary><b>Dashboard Discovery</b> (1)</summary>
 
 | Tool | Description |
 |------|-------------|
 | `list_dashboards` | List dashboards (by data app, workspace, or globally) |
-| `setup_embed_interactive` | Guided step-by-step flow for the full embed setup |
 </details>
 
 ---
@@ -320,7 +312,7 @@ This usually means no workspace was selected. Ask the assistant: *"List my works
 
 1. Check that the guest token is generated **server-side** (never expose API tokens in frontend code)
 2. Verify the embed ID and dashboard ID match
-3. Ask: *"Run a health check on my embed"* — this runs `diagnose_embed_health`
+3. Ask: *"Check my embed configuration for issues"* — the assistant inspects via `get_embed_details`
 
 ### Theme changes not showing
 

package/dist/index.js

@@ -6484,7 +6484,7 @@ tenant. For example, filter by department or user role within an organization.
 // src/resources/content/use-case-guides.ts
 var useCaseGuidesContent = `# Databrain Use Case Guides
 
-Databrain supports 4 core use cases for embedding analytics. Each maps to a specific MCP prompt and tool sequence.
+Databrain supports 3 core use cases for embedding analytics. Each maps to a specific MCP prompt and tool sequence.
 
 ---
 
@@ -6501,15 +6501,7 @@ Databrain supports 4 core use cases for embedding analytics. Each maps to a spec
 **Tool sequence:** \`list_data_apps\` \u2192 \`list_datamarts\` \u2192 \`create_embed\` with \`clientId\` (per client) \u2192 \`generate_guest_token\` \u2192 \`generate_embed_code\`
 **Key concepts:** \`clientId\` for tenant isolation, \`templateDashboardId\` for cloning, \`tableTenancySettings\` for row-level security.
 
-## Use Case 3: Enable Metric Creation
-**When to use:** You want end-users to build their own metrics/charts on embedded dashboards.
-**Prompt:** \`enable-metric-creation\`
-**Two modes:**
-- **DRAG_DROP:** Visual metric builder \u2014 users drag columns, pick chart types. No AI needed.
-- **CHAT:** AI-powered natural language \u2014 users type questions. Requires LLM connector in Databrain admin.
-**Tool sequence:** \`list_embeds\` \u2192 \`update_embed\` (set accessSettings and options) \u2192 \`generate_guest_token\` (with permissions)
-
-## Use Case 4: Explore and Query Data
+## Use Case 3: Explore and Query Data
 **When to use:** You want to explore available data and ask questions using natural language.
 **Prompt:** \`query-data\` (or \`populate-semantic-layer\` if starting from an empty semantic layer)
 **Path:** Exploration + querying (datasources, datamarts, and workspaces are created in the Databrain UI). The semantic layer is populated via MCP tools.
@@ -6520,17 +6512,17 @@ Databrain supports 4 core use cases for embedding analytics. Each maps to a spec
 
 ## Customization (After Any Setup)
 
-Once an embed exists, use these prompts to customize:
+Once an embed exists, use the \`brand-my-embed\` prompt for theming. For everything else, describe what you need and the assistant uses tools directly:
 
-| Prompt | Purpose |
-|--------|---------|
-| \`brand-my-embed\` | Colors, fonts, chart styles, theme presets |
-| \`setup-filters\` | String, date, number, SQL, and server-side filters |
-| \`localize-embed\` | Multi-language support and locale settings |
-| \`secure-my-embed\` | Permissions, access controls, role-based tokens |
-| \`manage-embeds\` | List, compare, and manage existing embeds |
-| \`query-data\` | Natural language querying and semantic layer management |
-| \`populate-semantic-layer\` | Auto-generate table/column descriptions for AI features |
+| Task | How |
+|------|-----|
+| Colors, fonts, chart styles | Prompt: \`brand-my-embed\` |
+| Filters (string, date, SQL) | \`update_embed\` with \`options.globalFilterOptions\` via body passthrough |
+| Localization / translations | \`update_embed\` with \`componentAttributes\` via body passthrough |
+| Permissions / access controls | \`update_embed\` + \`generate_guest_token\` with permissions |
+| Metric creation modes | \`update_embed\` accessSettings + \`generate_guest_token\` |
+| Manage / audit embeds | \`list_embeds\` \u2192 \`get_embed_details\` |
+| Semantic layer | Prompt: \`populate-semantic-layer\` or \`get_semantic_layer\` / \`update_semantic_layer\` |
 
 ---
 
@@ -6540,9 +6532,8 @@ Once an embed exists, use these prompts to customize:
 |----------|-------------------|
 | Do you have dashboards in Databrain UI? | Yes \u2192 Use Case 1 |
 | Does each client need their own dashboard? | Yes \u2192 Use Case 2 |
-| Do you want end-users to create metrics? | Yes \u2192 Use Case 3 |
-| Want to explore or query your data? | Yes \u2192 Use Case 4 |
-| Just need to brand/filter/secure? | Use customization prompts |
+| Want to explore or query your data? | Yes \u2192 Use Case 3 |
+| Just need to brand/customize? | Use \`brand-my-embed\` prompt or describe what you need |
 `;
 
 // src/resources/content/permission-schema.ts
@@ -6665,7 +6656,7 @@ var resources = [
   {
     uri: "databrain://use-case-guides",
     name: "Use Case Guides",
-    description: "Summary of 4 core use cases with tool sequences, prompt names, and decision guide",
+    description: "Summary of 3 core use cases with tool sequences, prompt names, and decision guide",
     content: useCaseGuidesContent
   },
   {
@@ -6747,7 +6738,7 @@ Only after they answer, call \`generate_embed_code\` with the \`embedId\`.
 - Prefer existing embeds when the user already has one for their target dashboard.
 - NEVER ask for service or API tokens \u2014 both are handled automatically.
 - \`workspaceName\` comes from dashboard data, not from the data app name.
-- After setup, suggest: \`brand-my-embed\`, \`setup-filters\`, \`localize-embed\`, \`secure-my-embed\`.`;
+- After setup, suggest: \`brand-my-embed\` for theming. For filters, localization, or permissions, just describe what you need and use \`update_embed\` / \`generate_guest_token\` directly.`;
 
 // src/prompts/embed-blank-dashboard.ts
 var embedBlankDashboardPrompt = `You are guiding a multi-tenant setup where each client gets their own blank (or template-based) dashboard. This is ideal for SaaS platforms where each customer sees their own analytics.
@@ -6811,83 +6802,7 @@ The frontend code is the same for all clients \u2014 the \`clientId\` in the gue
 - STOP after each step. Present results and wait for user input before proceeding.
 - When calling \`create_embed\` with \`clientId\`, only pass \`accessSettings: { datamartName }\` \u2014 the tool fills in all required defaults automatically. Do NOT enumerate isAllow* fields.
 - \`clientId\` must be consistent between embed creation and guest token generation.
-- After setup, suggest: \`brand-my-embed\`, \`setup-filters\`, \`enable-metric-creation\`, \`secure-my-embed\`.`;
-
-// src/prompts/enable-metric-creation.ts
-var enableMetricCreationPrompt = `You are helping the user enable end-user metric creation on their embedded dashboards. Databrain supports two modes: DRAG_DROP (visual builder) and CHAT (AI-powered natural language).
-
-**CRITICAL: This is an interactive guide. You MUST stop and ask the user at each decision point before proceeding. Never assume answers.**
-
-**Two Metric Creation Modes:**
-- **DRAG_DROP**: Visual metric builder where users drag columns, choose chart types, and configure metrics. No AI needed.
-- **CHAT**: AI-powered natural language metric creation. Users type questions like "Show me revenue by month" and Databrain generates the metric. Requires an LLM connector configured in the Databrain admin panel.
-
-**Step 1 \u2014 Identify the Embed**
-Call \`list_embeds\` to show available embeds.
-STOP and ASK the user: "Which embed do you want to enable metric creation on?"
-Only after they answer, call \`get_embed_details\` to see current access settings.
-
-**Step 2 \u2014 Choose the Mode**
-STOP and ASK the user:
-- "Which metric creation mode do you want?"
-  - **DRAG_DROP**: Visual builder, no AI dependency
-  - **CHAT**: AI-powered, requires LLM connector setup in Databrain admin
-Do NOT proceed until the user picks a mode.
-
-If they choose CHAT, tell them: "CHAT mode requires an LLM connector configured in the Databrain admin panel (Settings > AI Configuration). Have you set this up?"
-
-**Step 3 \u2014 Update Embed Access Settings**
-Call \`update_embed\` with the embed ID and updated \`accessSettings\`:
-
-For DRAG_DROP mode:
-\`\`\`json
-{
-  "accessSettings": {
-    "isAllowMetricCreation": true,
-    "metricCreationMode": "DRAG_DROP"
-  }
-}
-\`\`\`
-
-For CHAT mode:
-\`\`\`json
-{
-  "accessSettings": {
-    "isAllowMetricCreation": true,
-    "metricCreationMode": "CHAT"
-  }
-}
-\`\`\`
-
-**Step 4 \u2014 Configure Embed Options**
-Call \`update_embed\` (body passthrough with options) to ensure the UI supports metric creation:
-- \`disableMetricCreation\`: false
-- \`hideCreateMetricButton\`: false
-- \`hideChartSettings\`: false (so users can customize their created metrics)
-
-**Step 5 \u2014 Set Guest Token Permissions**
-Remind the user: guest token \`permissions\` override embed \`accessSettings\` at runtime. When generating guest tokens via \`generate_guest_token\`, include:
-\`\`\`json
-{
-  "permissions": {
-    "isDisableMetricCreation": false
-  }
-}
-\`\`\`
-
-**Step 6 \u2014 Verify and Test**
-Call \`get_embed_details\` to verify the settings were applied correctly.
-Tell the user to refresh their embed and look for the "+ Create Metric" button.
-
-If using CHAT mode, remind the user:
-- An LLM connector must be configured in Databrain admin (Settings > AI Configuration)
-- The datamart should have a populated semantic layer for best AI results (use \`get_semantic_layer\` to check, and \`update_semantic_layer\` or the \`populate-semantic-layer\` prompt to populate it)
-
-**Important rules:**
-- STOP and ASK at each decision point. Never skip ahead.
-- Guest token \`permissions\` take precedence over embed \`accessSettings\` \u2014 both must allow metric creation.
-- CHAT mode requires external LLM connector setup (not available via MCP).
-- After enabling, suggest: \`secure-my-embed\` to fine-tune which users get metric creation access.`;
+- After setup, suggest: \`brand-my-embed\` for theming. For filters, metric creation, or permissions, just describe what you need and use \`update_embed\` / \`generate_guest_token\` directly.`;
 
 // src/prompts/brand-my-embed.ts
 var brandMyEmbedPrompt = `You are helping the user brand and visually customize their embedded Databrain dashboard. Follow these steps.
@@ -6964,327 +6879,6 @@ Iterate by calling the relevant tools again with adjusted values. Remember: all
 
 **Reference**: Use \`databrain://theme-reference\` for the full theme schema and \`databrain://web-component-reference\` for all component attributes.`;
 
-// src/prompts/localize-embed.ts
-var localizeEmbedPrompt = `You are helping the user set up multi-language support for their embedded Databrain dashboard. Follow these steps.
-
-**Step 1 \u2014 Identify the Embed and Target Languages**
-Call \`list_embeds\` and ask the user which embed to localize. Then ask:
-- Which languages do they need? (Supported: English, French, Spanish, German, Italian, Portuguese, Chinese, Japanese, Korean, Hindi, Arabic, Russian)
-- What is the default/primary language?
-
-**Step 2 \u2014 Set the Primary Language**
-Use \`update_embed\` (body passthrough with componentAttributes) with the \`language\` parameter set to their primary language code (e.g. "fr" for French).
-
-**Step 3 \u2014 Gather Translations**
-Call \`get_embed_details\` to understand what text is in the dashboard. Then help the user build translations:
-
-Ask the user to list the key text elements in their dashboard:
-- Dashboard title
-- Metric titles and descriptions
-- Filter labels
-- Any footnotes or annotations
-
-For each text element, ask for translations in their target languages. Build translation entries like:
-\`\`\`
-{ key: "Revenue", translations: { fr: "Revenus", es: "Ingresos", de: "Einnahmen" } }
-\`\`\`
-
-**Important rules for translation keys:**
-- Keys must match the dashboard text EXACTLY (case-sensitive)
-- Define longer keys before shorter ones to avoid partial matches
-- If a translation is missing for a language, the original key text is shown as fallback
-
-**Step 4 \u2014 Apply Translations**
-Use \`update_embed\` (body passthrough with componentAttributes.translationDictionary) with the translations.
-
-**Step 5 \u2014 Configure Calendar (if needed)**
-Ask if they need a different calendar system:
-- \`"default"\` \u2014 standard Gregorian calendar
-- \`"ind"\` \u2014 Indian calendar format
-
-Set via \`calendarType\` in \`update_embed\` (body passthrough with componentAttributes).
-
-**Step 6 \u2014 Set Custom Messages**
-Ask if they want to translate system messages:
-- Token expiry message (shown when session expires)
-- Token absent message (shown when no auth token)
-
-Use \`update_embed\` (body passthrough with options.customMessages) with \`tokenExpiry\` and \`tokenAbsent\` in their primary language.
-
-**Step 7 \u2014 Test**
-Tell the user to refresh the embed and verify translations appear correctly. If any text isn't translating, double-check that the key matches exactly.
-
-Remind the user: translations can also be applied per-user at the web component level using the \`translation-dictionary\` attribute, which is useful for dynamically switching languages based on user preferences.
-
-**Tip:** The \`language\`, \`calendarType\`, \`hideDashboardFilters\`, and filter settings are all configurable via \`update_embed\` body passthrough.
-
-**Important:** After changing language or translation settings, existing guest tokens still carry the old configuration. Remind the user to **regenerate guest tokens** (or reload their app) for changes to take effect.`;
-
-// src/prompts/secure-my-embed.ts
-var secureMyEmbedPrompt = `You are helping the user secure and configure permissions for their embedded Databrain dashboard. Follow these steps.
-
-**Step 1 \u2014 Understand User Roles**
-Ask the user about their end-user roles. Common patterns:
-- **Viewer** \u2014 can see data, maybe filter, but can't modify anything
-- **Analyst** \u2014 can export data, drill down, use filters
-- **Builder/Power User** \u2014 can create/edit metrics, change layouts
-- **Admin** \u2014 full access including AI features
-
-Ask: "What roles do your end-users have? What should each role be allowed to do?"
-
-**Step 2 \u2014 Lock Down the Embed Defaults**
-Use \`update_embed\` (body passthrough with options) to set restrictive defaults (principle of least privilege):
-
-\`\`\`
-update_embed(embedId, body: { options: {
-  disableMetricCreation: true,
-  disableMetricUpdation: true,
-  disableMetricDeletion: true,
-  disableManageMetrics: true,
-  disableLayoutCustomization: true,
-  disableSaveLayout: true,
-  disableUnderlyingData: true,
-  disableScheduleEmailReports: true,
-  enableDownloadCsv: false,
-  enableDownloadAllPdf: false)
-\`\`\`
-
-Explain: "We start with everything locked down, then use guest token permissions to grant access per role."
-
-**Step 3 \u2014 Create Role-Based Guest Token Examples**
-Show the user how to generate tokens for each role:
-
-**Viewer (default \u2014 no overrides needed):**
-\`\`\`
-generate_guest_token(clientId, dataAppName)
-\`\`\`
-
-**Analyst (can export and drill down):**
-\`\`\`
-generate_guest_token(clientId, dataAppName, {
-  permissions: {
-    isEnableDownloadMetrics: true,
-    isEnableUnderlyingData: true
-  }
-})
-\`\`\`
-
-**Builder (can create and edit):**
-\`\`\`
-generate_guest_token(clientId, dataAppName, {
-  permissions: {
-    isEnableManageMetrics: true,
-    isEnableMetricUpdation: true,
-    isEnableCustomizeLayout: true
-  }
-})
-\`\`\`
-
-**Admin (full access with AI):**
-\`\`\`
-generate_guest_token(clientId, dataAppName, {
-  permissions: {
-    isEnableManageMetrics: true,
-    isEnableMetricUpdation: true,
-    isEnableCustomizeLayout: true,
-    isEnableDownloadMetrics: true,
-    isEnableUnderlyingData: true
-  }
-})
-\`\`\`
-
-**Step 4 \u2014 Multi-Tenancy Check**
-Ask: "Do different customers/organizations see different data?"
-
-If yes:
-- Explain that \`clientId\` in the guest token controls row-level security
-- The \`clientId\` must match values in the datamart's client column
-- Suggest using \`params.appFilters\` / \`params.dashboardAppFilters\` in \`generate_guest_token\` for server-side row-level filtering
-
-**Step 5 \u2014 Security Checklist**
-Walk through this checklist with the user:
-
-1. \u2705 API token is stored server-side only (never in frontend code)
-2. \u2705 Guest tokens are generated server-side per request
-3. \u2705 Guest tokens have appropriate expiry time
-4. \u2705 Embed defaults are restrictive (locked down in Step 2)
-5. \u2705 Role-based permissions are granted via guest token overrides
-6. \u2705 Multi-tenancy is configured if serving multiple customers
-7. \u2705 Download restrictions are set (\`disableDownloadDataNoFilters\` if needed)
-
-**Step 6 \u2014 Generate Secure Frontend Code**
-Call \`generate_embed_code\` for their framework. Emphasize that the generated code includes a backend route for token generation \u2014 this is the secure pattern.
-
-When reviewing runtime behavior (for example missing \`+ Create Metric\`), verify contract shape explicitly:
-- Guest token request body must send \`permissions\` at top level (not nested inside \`params\`)
-- Include \`dataAppName\` at top level
-- For metric creation flows, ensure \`permissions.isDisableMetricCreation=false\` and \`permissions.isEnableManageMetrics=true\`
-- Confirm embed options do not set \`disableMetricCreation=true\`
-- In browser network response for \`/api/v2/dashboard\`, confirm \`isAllowedToCreateMetrics=true\` and \`metricCreationMode=DRAG_DROP\`
-
-**Important:** After changing embed permissions or access settings, existing guest tokens still carry the old configuration. Remind the user to **regenerate guest tokens** (or reload their app) for changes to take effect.
-
-**Reference**: Use \`databrain://self-serve-reference\` for all permission flags and \`databrain://multi-tenancy-guide\` for row-level security details.`;
-
-// src/prompts/setup-filters.ts
-var setupFiltersPrompt = `You are helping the user configure dashboard filters for their embedded Databrain dashboard. Follow these steps.
-
-**Step 1 \u2014 Understand the Data**
-Call \`list_embeds\` and ask which embed to configure. Then call \`get_embed_details\` to see the current filter configuration.
-
-**Step 2 \u2014 Identify Filter Needs**
-Ask the user:
-- "What filters do your end-users need?" (e.g. date range, region, product category, status)
-- "Should any filters have default values?"
-- "Are there filters you want to hide from end-users?" (e.g. internal IDs)
-- "Do any filters have very large option lists?" (100+ options \u2014 these benefit from SQL-based filters)
-
-**Step 3 \u2014 Configure String Filters**
-For text/category filters, use \`update_embed\` (body passthrough with options.globalFilterOptions) for \`stringFilters\`:
-
-\`\`\`
-stringFilters: [{
-  filterName: "Region",
-  options: [
-    { value: "US", label: "United States" },
-    { value: "EU", label: "Europe" },
-    { value: "APAC", label: "Asia Pacific" }
-  ],
-  defaultOption: "US"
-}]
-\`\`\`
-
-For multi-select defaults, use an array: \`defaultOption: ["US", "EU"]\`
-
-**Step 4 \u2014 Configure Date Filters**
-For date range filters, use \`dateFilters\`:
-
-\`\`\`
-dateFilters: [{
-  filterName: "Date Range",
-  startDate: "2024-01-01",
-  endDate: "2024-12-31",
-  datePresetOptions: [
-    { label: "Last 30 Days", type: "Last", interval: "Days" },
-    { label: "This Quarter", type: "This", interval: "Quarter" },
-    { label: "This Year", type: "This", interval: "Year" }
-  ]
-}]
-\`\`\`
-
-**Step 5 \u2014 Configure Number Filters**
-For numeric range filters, use \`numberFilters\`:
-
-\`\`\`
-numberFilters: [{
-  filterName: "Revenue",
-  min: 0,
-  max: 1000000
-}]
-\`\`\`
-
-**Step 6 \u2014 SQL-Based Filters (for large datasets)**
-If a filter has 100+ options (e.g. company names, product SKUs), use \`sqlFilters\`:
-
-\`\`\`
-sqlFilters: [{
-  filterName: "Company",
-  sql: "SELECT DISTINCT \\"company_name\\" FROM \\"companies\\" ORDER BY \\"company_name\\"",
-  columnName: "company_name"
-}]
-\`\`\`
-
-Explain: "SQL filters dynamically fetch options from the database, so they're always up-to-date and don't bloat the embed configuration."
-
-**Step 7 \u2014 Hide Unnecessary Filters**
-If the user has filters that should be hidden (e.g. internal IDs, tenant columns already handled by clientId):
-
-\`\`\`
-hideDashboardFilters: ["internal_id", "tenant_id"]
-\`\`\`
-
-**Step 8 \u2014 Configure Date Picker Options**
-Ask if they want to hide certain date picker presets:
-
-\`\`\`
-hideDatePickerOptions: ["yesterday", "last week"]
-\`\`\`
-
-**Step 9 \u2014 Metric Filter Position**
-Ask: "Do you want metric-level filters inside the card or outside?"
-- \`"inside"\` \u2014 compact, within each metric card (default)
-- \`"outside"\` \u2014 above the card, more prominent
-
-Also ask if they want cross-metric filtering (\`enableMultiMetricFilters\`).
-
-**Step 10 \u2014 Server-Side Filters (via Guest Token)**
-If the user needs filters that end-users cannot remove (e.g. for security):
-- Explain that \`appFilters\` in \`generate_guest_token\` are server-side enforced
-- These are invisible to the end-user and cannot be overridden
-- Perfect for tenant filtering, role-based data access, or compliance requirements
-
-Show an example:
-\`\`\`
-generate_guest_token(clientId, dataAppName, {
-  params: {
-    appFilters: [{
-      metricId: "metric_123",
-      values: {
-        region: ["US", "CA"]
-      }
-    }]
-  }
-})
-\`\`\`
-
-**Important:** After changing filter settings, existing guest tokens still carry the old configuration. Remind the user to **regenerate guest tokens** (or reload their app) for changes to take effect.
-
-**Reference**: Use \`databrain://filter-reference\` for all filter types, operators, and combination rules.`;
-
-// src/prompts/manage-embeds.ts
-var manageEmbedsPrompt = `You are helping the user manage Databrain embed configurations \u2014 list, compare, and review settings. Follow these steps based on what the user needs.
-
-**Step 1 \u2014 Understand the Task**
-Ask the user what they want to do:
-1. **List and compare embeds** \u2014 see all embeds and their configurations
-2. **Review embed settings** \u2014 audit theme, permissions, and options
-
-**For Listing and Comparing Embeds:**
-1. Call \`list_embeds\` to show all embed configurations
-2. For detailed comparison, call \`get_embed_details\` on each embed the user wants to compare
-3. Present a comparison table showing:
-   - Embed name and ID
-   - Linked dashboard
-   - Theme (preset or custom)
-   - Key permissions (metric creation, downloads, AI pilot)
-   - Multi-tenancy settings
-
-**For Reviewing Embed Settings:**
-1. Call \`get_embed_details\` for the embed
-2. Present a clear summary organized by category:
-
-   **Theme:**
-   - Current preset or custom theme values
-   - Suggest \`customize_embed_theme\` if they want changes
-
-   **Permissions / Access Settings:**
-   - What's enabled vs disabled
-   - Suggest \`update_embed\` for changes
-
-   **Filters:**
-   - Current filter configuration
-   - Suggest \`update_embed\` (body passthrough with options.globalFilterOptions) for changes
-
-   **Internationalization:**
-   - Current language and translations
-   - Suggest \`update_embed\` (body passthrough with componentAttributes) for changes
-
-3. Ask: "Would you like to modify any of these settings?"
-
-**Tips:**
-- Embed IDs are permanent \u2014 even if you rename an embed, the ID stays the same
-- After any embed setting changes, existing guest tokens still carry the old configuration \u2014 **regenerate guest tokens** for changes to take effect`;
-
 // src/prompts/query-data.ts
 var queryDataPrompt = `You are helping the user query their data using natural language. Databrain's AI converts questions into SQL, runs them against the database, and returns results with chart type suggestions.
 
@@ -7453,36 +7047,11 @@ var prompts = [
     description: "Multi-tenant blank dashboard setup: each client gets their own dashboard copy via create_embed with clientId",
     content: embedBlankDashboardPrompt
   },
-  {
-    name: "enable-metric-creation",
-    description: "Enable end-user metric creation on embeds: DRAG_DROP (visual) or CHAT (AI-powered) modes",
-    content: enableMetricCreationPrompt
-  },
   {
     name: "brand-my-embed",
     description: "Customize colors, fonts, chart styles, and component theme to match your brand",
     content: brandMyEmbedPrompt
   },
-  {
-    name: "localize-embed",
-    description: "Set up multi-language support with translations and locale settings",
-    content: localizeEmbedPrompt
-  },
-  {
-    name: "secure-my-embed",
-    description: "Configure permissions, access controls, and role-based guest tokens",
-    content: secureMyEmbedPrompt
-  },
-  {
-    name: "setup-filters",
-    description: "Configure dashboard filters: string, date, number, SQL-based, and server-side filters",
-    content: setupFiltersPrompt
-  },
-  {
-    name: "manage-embeds",
-    description: "List, compare, rename, delete embeds and rotate API tokens",
-    content: manageEmbedsPrompt
-  },
   {
     name: "query-data",
     description: "Query your data with natural language: ask questions, get results with chart suggestions, manage the semantic layer",
@@ -7523,16 +7092,15 @@ function createServer(config) {
 \u2022 SERVICE TOKEN (DATABRAIN_SERVICE_TOKEN): org-level, pre-configured in MCP server config. Used for: list_datasources, discover_datasource_schema, list_datamarts, list_workspaces, link_datamart_to_workspace, list_data_apps, create_data_app, list_api_tokens, create_api_token, rotate_api_token, list_dashboards (workspace scope), ask_question, get_semantic_layer, update_semantic_layer. You already have this \u2014 never ask the user for it.
 \u2022 API TOKEN (per data app): scoped to one data app, used for embedding operations (list_embeds, get_embed_details, create_embed, update_embed, generate_guest_token) and data-app scoped dashboard listing. Created/rotated via create_api_token or rotate_api_token during setup.
 
-4 core use cases \u2014 match the user's scenario to the right prompt:
+3 core use cases \u2014 match the user's scenario to the right prompt:
 1. Embed existing dashboard \u2192 prompt: embed-existing-dashboard (list_data_apps \u2192 create_api_token \u2192 list_dashboards \u2192 create_embed \u2192 generate_guest_token \u2192 generate_embed_code)
 2. Blank dashboard per client \u2192 prompt: embed-blank-dashboard (list_data_apps \u2192 list_datamarts \u2192 create_embed with clientId per client \u2192 generate_guest_token \u2192 generate_embed_code)
-3. Enable metric creation \u2192 prompt: enable-metric-creation (list_embeds \u2192 update_embed accessSettings \u2192 generate_guest_token with permissions)
-4. Explore & query data \u2192 prompt: query-data (list_datamarts \u2192 get_semantic_layer \u2192 if empty, auto-generate descriptions with update_semantic_layer or use populate-semantic-layer prompt \u2192 ask_question)
+3. Explore & query data \u2192 prompt: query-data (list_datamarts \u2192 get_semantic_layer \u2192 if empty, auto-generate descriptions with update_semantic_layer or use populate-semantic-layer prompt \u2192 ask_question)
 
 Infrastructure CREATION (datasources, datamarts, workspaces) is done in the Databrain UI.
 However, the **semantic layer** is populated and maintained entirely via MCP tools (get_semantic_layer, update_semantic_layer). When the semantic layer is empty, YOU (the assistant) should auto-generate descriptions from table/column names and push them with update_semantic_layer \u2014 do NOT send the user to the UI for this.
 
-After embed setup, customize with: brand-my-embed, setup-filters, localize-embed, secure-my-embed.
+After embed setup, customize with: brand-my-embed prompt for theming. Use update_embed (body passthrough) for filters, localization, and permissions.
 Use generate_guest_token for frontend auth and generate_embed_code for framework snippets.
 For advanced embed configuration (UI toggles, filters, i18n), use update_embed with a body passthrough.
 To diagnose embed issues, use get_embed_details and reason about the response.
@@ -7566,7 +7134,7 @@ async function main() {
     const server = createServer(config);
     const transport = new StdioServerTransport();
     await server.connect(transport);
-    banner(`v${VERSION} ready (24 tools, 11 resources, 10 prompts)`);
+    banner(`v${VERSION} ready (21 tools, 11 resources, 5 prompts)`);
   } catch (error) {
     logger.error("Failed to start server", error);
     process.exit(1);