@kadoa/mcp 0.7.0-rc.2 → 0.7.1

package/dist/index.js

@@ -17,7 +17,7 @@ var package_default;
 var init_package = __esm(() => {
   package_default = {
     name: "@kadoa/mcp",
-    version: "0.7.0-rc.2",
+    version: "0.7.1",
     description: "Kadoa MCP Server — manage workflows from Claude Desktop, Cursor, and other MCP clients",
     type: "module",
     main: "dist/index.js",
@@ -44,7 +44,7 @@ var init_package = __esm(() => {
       prepublishOnly: "bun run check-types && bun run test:unit && bun run build"
     },
     dependencies: {
-      "@kadoa/node-sdk": "^0.45.0",
+      "@kadoa/node-sdk": "^0.48.0",
       "@modelcontextprotocol/client": "2.1.0",
       "@modelcontextprotocol/express": "2.0.1",
       "@modelcontextprotocol/node": "2.1.0",
@@ -104,6 +104,11 @@ function createKadoaClient(auth) {
 var MCP_SCRAPE_REQUEST_TIMEOUT_MS = 310000;
 var init_client = () => {};
 
+// src/default-product.ts
+function parseDefaultProductType(value) {
+  return value === "realtime" || value === "workflow" ? value : undefined;
+}
+
 // src/error-summary.ts
 function redactCredentialText(value) {
   return value.replace(/Bearer\s+[^\s,;]+/gi, "Bearer [redacted]").replace(/\b(?:kcs_|tk-)[A-Za-z0-9._~-]+\b/g, "[redacted]").replace(/\beyJ[A-Za-z0-9_-]+\.[A-Za-z0-9_-]+\.[A-Za-z0-9_-]+\b/g, "[redacted]");
@@ -209,7 +214,7 @@ function withVariableDashboardUrl(variable) {
 
 // src/tools/shared/output-schemas.ts
 import { z } from "zod";
-var apiField, nullableString, PaginationOutput, WorkflowRowOutput, ListWorkflowsOutput, GetWorkflowOutput, FetchDataOutput, ASSISTANT_ACTIVITIES, WorkflowAssistantOutput, AssistantTimelineOutput, AssistantAnswerOutput, AssistantControlOutput, WorkflowStrategyOutput;
+var apiField, nullableString, PaginationOutput, workflowStatusFields, WorkflowRowOutput, ListWorkflowsOutput, GetWorkflowOutput, FetchDataOutput, ASSISTANT_ACTIVITIES, WorkflowAssistantOutput, AssistantTimelineOutput, AssistantAnswerOutput, AssistantControlOutput, WorkflowStrategyOutput;
 var init_output_schemas = __esm(() => {
   apiField = z.unknown().optional();
   nullableString = z.string().nullable();
@@ -224,11 +229,20 @@ var init_output_schemas = __esm(() => {
     nextCursor: nullableString.optional(),
     next: z.record(z.string(), z.unknown()).optional()
   }).loose();
+  workflowStatusFields = {
+    status: z.string(),
+    badges: z.array(z.object({ kind: z.string(), status: z.string(), label: z.string() }).loose()),
+    group: nullableString,
+    reviewReason: z.object({ code: z.string(), explanation: z.string() }).loose().nullable(),
+    execution: apiField,
+    assistantState: apiField,
+    workspaceActivity: apiField
+  };
   WorkflowRowOutput = z.object({
     id: apiField,
     name: apiField,
     dashboardUrl: z.string(),
-    status: z.string(),
+    ...workflowStatusFields,
     state: nullableString,
     displayState: nullableString,
     runState: nullableString,
@@ -239,7 +253,6 @@ var init_output_schemas = __esm(() => {
     createdAt: apiField,
     assistant: z.object({
       eligible: z.boolean(),
-      dataStale: z.boolean(),
       session: z.object({ id: z.string(), threadId: apiField, status: apiField }).loose().nullable()
     }).loose(),
     template: z.object({ id: z.string(), pinnedVersion: apiField, latestVersion: apiField, isOutdated: z.boolean() }).loose().nullable(),
@@ -252,7 +265,7 @@ var init_output_schemas = __esm(() => {
     name: apiField,
     dashboardUrl: z.string(),
     description: apiField,
-    status: z.string(),
+    ...workflowStatusFields,
     state: nullableString,
     displayState: nullableString,
     runState: nullableString,
@@ -263,7 +276,6 @@ var init_output_schemas = __esm(() => {
     entity: apiField,
     schema: apiField,
     navigationMode: apiField,
-    dataStale: apiField,
     assistantEligible: z.boolean(),
     assistantSessionId: apiField,
     assistantThreadId: apiField,
@@ -292,8 +304,6 @@ var init_output_schemas = __esm(() => {
     createdAt: apiField,
     notificationConfig: apiField,
     channelCounts: apiField,
-    dataValidation: apiField,
-    validationStatistics: apiField,
     realtime: z.object({ lastDataChangedAt: apiField, observerHealth: apiField }).loose().nullable()
   }).loose();
   FetchDataOutput = z.object({
@@ -1888,11 +1898,11 @@ function scrapeResult(response, format) {
   };
   return withExtraContent(textOnlyResult(meta), { type: "text", text });
 }
-function registerScrapeTools(server, ctx, helpers) {
+function registerScrapeTools(server, ctx, helpers, defaultProductType) {
   const { withErrorHandling } = helpers;
   server.registerTool("scrape", {
     title: "Scrape URL",
-    description: "Immediately fetch one URL as LLM-friendly markdown or raw HTML without creating a workflow. " + "Markdown is the default. Routing is fully automatic: it starts cheap and escalates through stealth tiers on its own when a site blocks bots, and remembers the working method per host. " + "Use create_workflow instead for structured extraction, recurring runs, monitoring, or navigation-heavy jobs. " + "The result is a JSON metadata block followed by the page text as its own block, capped at 70 KB.",
+    description: "Immediately fetch one URL as LLM-friendly markdown or raw HTML without creating a workflow. " + "Markdown is the default. Routing is fully automatic: it starts cheap and escalates through stealth tiers on its own when a site blocks bots, and remembers the working method per host. " + (defaultProductType === "realtime" ? "Use create_realtime_monitor for monitoring, and create_workflow for one-time or scheduled structured extraction or navigation-heavy jobs. " : "Use create_workflow instead for structured extraction, recurring runs, monitoring, or navigation-heavy jobs. ") + "The result is a JSON metadata block followed by the page text as its own block, capped at 70 KB.",
     inputSchema: strictSchema({
       url: z7.url().refine((value) => {
         const protocol = new URL(value).protocol;
@@ -2161,24 +2171,31 @@ function registerWhoamiTool(server, ctx, helpers) {
   const { withErrorHandling } = helpers;
   server.registerTool("whoami", {
     title: "Show Current User",
-    description: "Show current user details: email, authentication method, and team memberships",
+    description: "Show current user details: email, authentication method, team memberships, and the team's default product",
     inputSchema: strictSchema({}),
     annotations: { readOnlyHint: true, destructiveHint: false, idempotentHint: true, openWorldHint: false }
   }, withErrorHandling("whoami", async () => {
+    const defaultProductTypePromise = ctx.client.user.getFeatures().then((features) => parseDefaultProductType(features.defaultProductType), () => {
+      return;
+    });
     if (ctx.principal.connector) {
-      const teams = await ctx.client.listTeams();
+      const [teams, defaultProductType] = await Promise.all([
+        ctx.client.listTeams(),
+        defaultProductTypePromise
+      ]);
       const activeTeam = teams.find((team) => team.id === ctx.principal.teamId);
       if (!activeTeam) {
         throw new Error("The active team is unavailable for this credential. Reconnect and choose a team.");
       }
       return jsonResult({
         authMethod: "Shared organization service account",
-        team: { name: activeTeam.name }
+        team: { name: activeTeam.name, ...defaultProductType ? { defaultProductType } : {} }
       });
     }
-    const [user, teams] = await Promise.all([
+    const [user, teams, defaultProductType] = await Promise.all([
       ctx.client.user.getCurrentUser(),
-      ctx.client.listTeams()
+      ctx.client.listTeams(),
+      defaultProductTypePromise
     ]);
     const activeTeam = teams.find((team) => team.id === ctx.principal.teamId);
     if (!activeTeam) {
@@ -2189,7 +2206,8 @@ function registerWhoamiTool(server, ctx, helpers) {
       authMethod: ctx.principal.actor ? `Kadoa Assistant (delegated via ${ctx.principal.actor.sub})` : "Kadoa access token",
       team: {
         name: activeTeam.name,
-        ...activeTeam.memberRole ? { memberRole: activeTeam.memberRole } : {}
+        ...activeTeam.memberRole ? { memberRole: activeTeam.memberRole } : {},
+        ...defaultProductType ? { defaultProductType } : {}
       },
       scopes: ctx.principal.scopes
     });
@@ -2649,49 +2667,109 @@ var init_variables = __esm(() => {
   init_schemas();
 });
 
-// src/tools/workflows.ts
-import { WorkflowStatusFilter } from "@kadoa/node-sdk";
-import { z as z12 } from "zod";
+// src/tools/shared/workflow-status.ts
+function reviewReasonExplanation(reason) {
+  return reason && REVIEW_REASON_EXPLANATIONS[reason] || UNKNOWN_REVIEW_REASON;
+}
 function isRealTimeInterval(interval) {
   return interval?.toUpperCase().replace("-", "_") === "REAL_TIME";
 }
-function deriveStatusLabel(workflow) {
-  const ds = workflow.displayState ?? workflow.state;
-  switch (ds) {
-    case "ACTIVE":
-      if (isRealTimeInterval(workflow.updateInterval))
-        return "Running";
-      if (!workflow.updateInterval || workflow.updateInterval === "only once")
-        return "Complete";
-      return "Scheduled";
-    case "RUNNING":
-      return "Running";
-    case "VALIDATING":
-      return "Validating";
-    case "FAILED":
-      return "Failed";
-    case "PAUSED":
-      return "Paused";
-    case "SETUP":
-      return "Sample Processing";
-    case "QUEUED":
-      return "Queued";
-    case "PENDING_START":
-      return "Not Started";
-    case "PREVIEW":
-      return "Pending Review";
-    case "DEGRADED":
-      return "Degraded";
-    case "DELETED":
-      return "Deleted";
-    case "COMPLIANCE_REVIEW":
-      return "Compliance Review";
-    case "COMPLIANCE_REJECTED":
-      return "Rejected";
-    default:
-      return ds ?? "Unknown";
+function fallbackStatusCode(w) {
+  const ds = w.displayState ?? w.state ?? null;
+  if (ds === "ACTIVE") {
+    if (w.isRealTime === true || isRealTimeInterval(w.updateInterval))
+      return "RUNNING";
+    return w.updateInterval && !["ONLY_ONCE", "only once"].includes(w.updateInterval) ? "SCHEDULED" : "COMPLETE";
   }
+  if (ds === "VALIDATING")
+    return "AWAITING_REVIEW";
+  return ds;
+}
+function labelFor(code, previewReady) {
+  const label = WORKFLOW_STATUS_LABELS[code === "PREVIEW" && previewReady ? "PREVIEW_READY" : code];
+  if (label)
+    return label;
+  console.error(`[workflow-status] unmapped workflow status, passing through verbatim: ${code}`);
+  return code;
+}
+function workflowStatusLabel(w) {
+  const code = w.presentation?.status ?? fallbackStatusCode(w);
+  if (!code)
+    return "Unknown";
+  return labelFor(code, w.availableActions?.approve?.allowed === true);
+}
+function workflowStatusView(w) {
+  const previewReady = w.availableActions?.approve?.allowed === true;
+  return {
+    status: workflowStatusLabel(w),
+    badges: (w.presentation?.badges ?? []).map((b) => ({
+      kind: b.kind,
+      status: b.status,
+      label: labelFor(b.status, previewReady)
+    })),
+    group: w.presentation?.group ?? null,
+    reviewReason: w.reviewReason ? { code: w.reviewReason, explanation: reviewReasonExplanation(w.reviewReason) } : null
+  };
 }
+function runStatusLabel(state) {
+  if (!state)
+    return "Unknown";
+  return RUN_STATE_LABELS[state] ?? state;
+}
+var REVIEW_STATUS_LABEL = "Verifying data", WORKFLOW_STATUS_LABELS, REVIEW_REASON_EXPLANATIONS, UNKNOWN_REVIEW_REASON = "Kadoa's quality or compliance checks selected this run for review.", RUN_STATE_LABELS;
+var init_workflow_status = __esm(() => {
+  WORKFLOW_STATUS_LABELS = {
+    ASSISTANT_WORKING: "Agent working",
+    AWAITING_REVIEW: REVIEW_STATUS_LABEL,
+    COMPLETE: "Complete",
+    COMPLIANCE_REJECTED: "Rejected",
+    COMPLIANCE_REVIEW: "Compliance review",
+    DEGRADED: "Degraded",
+    DELETED: "Deleted",
+    DRAFT: "Draft",
+    ERROR: "Failed",
+    FAILED: "Failed",
+    IN_SUPPORT: "In support",
+    NOT_SUPPORTED: "Not supported",
+    PAUSED: "Paused",
+    PENDING_START: "Not started",
+    PREVIEW: "Preview",
+    PREVIEW_READY: "Preview ready",
+    QUEUED: "Queued",
+    REPLY_NEEDED: "Reply needed",
+    RUNNING: "Running",
+    SCHEDULED: "Scheduled",
+    SETUP: "Setup",
+    STOPPED: "Stopped",
+    UPDATE_FAILED: "Update failed",
+    UPDATING: "Updating"
+  };
+  REVIEW_REASON_EXPLANATIONS = {
+    COMPLIANCE_REVIEW: "This run triggered a compliance check before delivery.",
+    HARD_FAIL: "An error stopped this run from completing normally.",
+    TRANSFORMATION_REPAIR_REQUIRED: "A data transformation could not be fixed automatically.",
+    MINIMUM_HUMAN_APPROVALS: "Kadoa's quality checks selected this run for review before delivery.",
+    DATA_QUALITY_SCHEMA_ADHERENCE: "The extracted data did not match this workflow's fields and data types.",
+    DATA_QUALITY_SCHEMA_VALIDATION: "Extracted values did not match this workflow's validation rules.",
+    DATA_QUALITY_COMPLETENESS: "The extracted record count did not match the expected source count.",
+    DATA_QUALITY_SEMANTIC_PLAUSIBILITY: "Some extracted values look unusual for their fields.",
+    ROW_COUNT_CHANGED: "This run's row count changed significantly.",
+    WORKFLOW_REVISION_DRIFT: "This run no longer matches the workflow's current setup.",
+    ZERO_RESULTS: "The source returned no data."
+  };
+  RUN_STATE_LABELS = {
+    FINISHED: "Success",
+    IN_PROGRESS: "Running",
+    VALIDATING: REVIEW_STATUS_LABEL,
+    FAILED: "Failed",
+    FAILED_INSUFFICIENT_FUNDS: "Failed",
+    NOT_SUPPORTED: "Failed"
+  };
+});
+
+// src/tools/workflows.ts
+import { WorkflowStatusFilter } from "@kadoa/node-sdk";
+import { z as z12 } from "zod";
 function mapRunStatus(state) {
   switch (state) {
     case "FINISHED":
@@ -2789,7 +2867,7 @@ function summarizeWorkflowAuditEntry(entry) {
 function historyEmptyHint(includeSnapshots) {
   return includeSnapshots ? "A single entry is too large to return inline. Retry without includeSnapshots for the diff summary." : "A single entry is too large to return inline even without snapshots. Retry with a narrower page (e.g. limit=1) or fewer fields.";
 }
-function registerWorkflowCoreTools(server, ctx, helpers) {
+function registerWorkflowCoreTools(server, ctx, helpers, defaultProductType) {
   const { withErrorHandling } = helpers;
   function resolveUrls(args) {
     const urls = args.urls ?? (args.url ? [args.url] : []);
@@ -2897,7 +2975,7 @@ function registerWorkflowCoreTools(server, ctx, helpers) {
 
 ` + "Create a data extraction workflow using agentic navigation. Supports one-time or scheduled runs. " + "If entity and schema are provided, they guide the extraction; otherwise the AI agent auto-detects the schema from the page. " + "The workflow runs asynchronously and may take several minutes. Do NOT poll or sleep-wait for completion. " + `Return the workflow ID to the user and let them check back later with get_workflow or fetch_data.
 
-` + "PREFER TEMPLATES: If the user's request matches an existing template, instantiate it via `templateId` instead of writing a fresh prompt/schema. " + "Use `list_templates` to discover available templates and `get_template` to inspect schemas before deciding. " + "When creating from a template, call this tool with the template's `templateId`, the source `urls`, optional `templateVersion`, and optional `prompt` containing only workflow-specific additional instructions. Never copy the template prompt, entity, or schema into the call, and never silently match or rewrite inline configuration.\n\n" + "NOTE: This tool is for one-time or scheduled extraction ONLY. " + "For continuous realtime monitoring that watches a page for data changes and sends alerts, use the create_realtime_monitor tool instead.",
+` + "PREFER TEMPLATES: If the user's request matches an existing template, instantiate it via `templateId` instead of writing a fresh prompt/schema. " + "Use `list_templates` to discover available templates and `get_template` to inspect schemas before deciding. " + "When creating from a template, call this tool with the template's `templateId`, the source `urls`, optional `templateVersion`, and optional `prompt` containing only workflow-specific additional instructions. Never copy the template prompt, entity, or schema into the call, and never silently match or rewrite inline configuration.\n\n" + "NOTE: This tool is for one-time or scheduled extraction ONLY. " + (defaultProductType === "realtime" ? `This team's default product is realtime monitoring. Do not call this tool unless the user explicitly asked for a one-time run (e.g. "once", "just the current list") or a schedule (e.g. "daily"). A request to get, extract, or scrape data that names no frequency is not explicit: call create_realtime_monitor instead.` : "For continuous realtime monitoring that watches a page for data changes and sends alerts, use the create_realtime_monitor tool instead."),
     inputSchema: strictSchema({
       prompt: extractionPromptInput.describe('Natural language description of what to extract (e.g., "Extract product prices and names"). Required unless templateId is provided. With templateId, use it only for workflow-specific instructions to add to the template prompt.'),
       ...extractionInputShape,
@@ -3050,7 +3128,7 @@ function registerWorkflowCoreTools(server, ctx, helpers) {
   }
   server.registerTool("create_realtime_monitor", {
     title: "Create Realtime Monitor",
-    description: "Create a realtime monitoring workflow that continuously watches a page for data changes and sends alerts. " + "Use this when the user wants to monitor, track, or watch a page in realtime. " + "At least one notification channel (email, webhook, slack, or websocket) is required; notification settings are persisted before Assistant creation is dispatched. " + `Realtime monitors CANNOT be converted to/from one-time or scheduled extraction workflows after creation.
+    description: "Create a realtime monitoring workflow that continuously watches a page for data changes and sends alerts. " + (defaultProductType === "realtime" ? "This team's default product is realtime monitoring: use this tool whenever the user asks to monitor, track, or watch a page, or to get or extract data from it, unless they explicitly ask for a one-time run or a schedule. A request that names no frequency counts as realtime. " : "Use this when the user wants to monitor, track, or watch a page in realtime. ") + "At least one notification channel (email, webhook, slack, or websocket) is required; notification settings are persisted before Assistant creation is dispatched. " + `Realtime monitors CANNOT be converted to/from one-time or scheduled extraction workflows after creation.
 
 ` + "If entity and schema are provided, they guide what the Assistant should monitor; otherwise the Assistant determines what to watch. " + "Creation is asynchronous and the response includes workflow, session, thread, and job IDs plus a dashboard URL. " + "Use generic workflow Assistant tools (get_workflow_assistant, answer_workflow_assistant_question, interrupt_workflow_assistant, resume_workflow_assistant, stop_workflow_assistant) for follow-up status, questions, and controls. " + "Do NOT poll or sleep-wait for completion.",
     inputSchema: strictSchema({
@@ -3107,7 +3185,7 @@ function registerWorkflowCoreTools(server, ctx, helpers) {
   }));
   server.registerTool("list_workflows", {
     title: "List Workflows",
-    description: `List workflows with customer-visible status, raw workflow and run states, Assistant session availability, template version drift, realtime freshness or health, and awaitingUserInput when the requesting user must act. Uses one paginated list request; use get_workflow only when you need full details for a selected workflow. To answer "what needs my attention", pass statusFilters: ["group:attention"]; this returns exactly the dashboard's "Needs your attention" group. In group:attention, a row with awaitingUserInput.prompt is an Assistant question: use get_workflow_assistant then answer_workflow_assistant_question. A row with awaitingUserInput but no prompt is a setup the user left unfinished: use get_workflow_assistant to see where it stopped, then resume_workflow_assistant or request_workflow_update. A row with displayState PREVIEW and no awaitingUserInput is preview data: use approve_workflow. Workflows with an open support ticket are excluded from group:attention and appear under group:support.`,
+    description: `List workflows with customer-visible status, raw workflow and run states, Assistant session availability, template version drift, realtime freshness or health, and awaitingUserInput when the requesting user must act. Uses one paginated list request; use get_workflow only when you need full details for a selected workflow. To answer "what needs my attention", pass statusFilters: ["group:attention"]; this returns exactly the dashboard's "Needs your attention" group. In group:attention, a row with awaitingUserInput.prompt is an Assistant question: use get_workflow_assistant then answer_workflow_assistant_question. A row with awaitingUserInput but no prompt is a setup the user left unfinished: use get_workflow_assistant to see where it stopped, then resume_workflow_assistant or request_workflow_update. A row with status "Preview ready" is preview data ready to approve with approve_workflow; "Preview" means preview data exists but approval isn't available yet. Workflows with an open support ticket are excluded from group:attention and appear under group:support. badges[].label is exactly what the Kadoa dashboard displays (lifecycle, activity, and support badges, e.g. "Setup" + "Agent working"): use these labels when describing the workflow to the user. status is the one-line summary in dashboard wording (e.g. "Verifying data" = the latest run is held for quality review before delivery; see reviewReason for why).`,
     inputSchema: strictSchema({
       limit: limitInput({
         noun: "Workflows",
@@ -3145,10 +3223,13 @@ function registerWorkflowCoreTools(server, ctx, helpers) {
         id: w.id,
         name: w.name,
         dashboardUrl: workflowDashboardUrl(w.id),
-        status: deriveStatusLabel(w),
+        ...workflowStatusView(w),
         state: w.state ?? null,
         displayState: w.displayState ?? null,
         runState: w.runState ?? null,
+        execution: w.execution ?? null,
+        assistantState: w.assistantState ?? null,
+        workspaceActivity: w.workspaceActivity ?? null,
         urls: w.urls,
         totalRecords: w.totalRecords,
         startedAt: w.startedAt,
@@ -3156,7 +3237,6 @@ function registerWorkflowCoreTools(server, ctx, helpers) {
         createdAt: w.createdAt,
         assistant: {
           eligible: w.assistantEligible === true,
-          dataStale: w.dataStale === true,
           session: w.assistantSessionId ? {
             id: w.assistantSessionId,
             threadId: w.assistantThreadId ?? null,
@@ -3188,7 +3268,7 @@ function registerWorkflowCoreTools(server, ctx, helpers) {
   }));
   server.registerTool("get_workflow", {
     title: "Get Workflow",
-    description: "Get canonical workflow details, including Assistant/session linkage, extraction intent, template-controlled parts, stale-data state, job/run state, schedule timezone, location, monitoring, validation, and realtime health. For Assistant-built workflows, extractionSpec - not the legacy prompt - is authoritative.",
+    description: 'Get canonical workflow details, including Assistant/session linkage, extraction intent, template-controlled parts, job/run state, schedule timezone, location, monitoring, and realtime health. For Assistant-built workflows, extractionSpec - not the legacy prompt - is authoritative. badges[].label is exactly what the Kadoa dashboard displays (lifecycle, activity, and support badges, e.g. "Setup" + "Agent working"): use these labels when describing the workflow to the user. status is the one-line summary in dashboard wording (e.g. "Verifying data" = the latest run is held for quality review before delivery; see reviewReason for why).',
     inputSchema: strictSchema({
       workflowId: z12.string().min(1).describe("The workflow ID")
     }),
@@ -3213,10 +3293,13 @@ function registerWorkflowCoreTools(server, ctx, helpers) {
       name: workflow.name,
       dashboardUrl: workflowDashboardUrl(workflow.id ?? args.workflowId),
       description: workflow.description,
-      status: deriveStatusLabel(workflow),
+      ...workflowStatusView(workflow),
       state: workflow.state ?? null,
       displayState: workflow.displayState ?? null,
       runState: workflow.runState ?? null,
+      execution: workflow.execution ?? null,
+      assistantState: workflow.assistantState ?? null,
+      workspaceActivity: workflow.workspaceActivity ?? null,
       urls: workflow.urls,
       prompt: extractionSpec ? null : legacyPrompt,
       intentSource: extractionSpec ? "extraction_spec" : legacyPrompt ? "legacy_prompt" : null,
@@ -3224,7 +3307,6 @@ function registerWorkflowCoreTools(server, ctx, helpers) {
       entity: workflow.entity,
       schema: workflow.schema,
       navigationMode: workflow.navigationMode,
-      dataStale: workflow.dataStale ?? false,
       assistantEligible: workflow.assistantEligible === true,
       assistantSessionId: workflow.assistantSessionId ?? null,
       assistantThreadId: workflow.assistantThreadId ?? null,
@@ -3267,8 +3349,6 @@ function registerWorkflowCoreTools(server, ctx, helpers) {
       createdAt: workflow.createdAt,
       notificationConfig: workflow.notificationConfig,
       channelCounts: workflow.channelCounts ?? {},
-      dataValidation: workflow.dataValidation ?? null,
-      validationStatistics: workflow.validationStatistics ?? null,
       realtime: isRealtime ? {
         lastDataChangedAt: workflow.lastDataChangedAt ?? null,
         observerHealth: workflow.observerHealth ?? null
@@ -3368,7 +3448,7 @@ function registerWorkflowHistoryTools(server, ctx, helpers) {
   }));
   server.registerTool("list_workflow_runs", {
     title: "List Workflow Runs",
-    description: "List a workflow's execution run history - each run's status, start/finish time, " + "record count, and errors. Use to answer 'when did this last succeed?' " + "(status='success', limit=1) or 'recent success/failure pattern?'. Distinct from " + "get_workflow_history, which is the config audit log. `status` is normally " + "success | failed | in_progress; a run whose backend state this server does not " + "recognize reports that raw state verbatim, so treat any other value as unknown " + "rather than as a failure.",
+    description: "List a workflow's execution run history - each run's status, start/finish time, " + "record count, and errors. Use to answer 'when did this last succeed?' " + "(status='success', limit=1) or 'recent success/failure pattern?'. Distinct from " + "get_workflow_history, which is the config audit log. `status` is normally " + "success | failed | in_progress; a run whose backend state this server does not " + "recognize reports that raw state verbatim, so treat any other value as unknown " + "rather than as a failure. `label` is the dashboard's wording for the run " + "(Success, Failed, Running, or 'Verifying data' = held for quality review before delivery); " + "prefer it when describing a run to the user.",
     inputSchema: strictSchema({
       workflowId: z12.string().min(1).describe("The workflow ID"),
       status: z12.enum(["success", "failed", "in_progress"]).optional().describe("Filter runs by outcome"),
@@ -3387,6 +3467,7 @@ function registerWorkflowHistoryTools(server, ctx, helpers) {
     const runs = (response.workflowRuns ?? []).map((r) => ({
       id: r.id,
       status: mapRunStatus(r.state),
+      label: runStatusLabel(r.state),
       startedAt: r.startedAt,
       finishedAt: r.finishedAt,
       records: r.records,
@@ -3436,7 +3517,7 @@ function registerWorkflowRunTools(server, ctx, helpers) {
   const FETCH_DATA_MAX_LIMIT = 500;
   server.registerTool("fetch_data", {
     title: "Fetch Workflow Data",
-    description: "Get a PAGE of extracted data from a workflow. Use ONLY for previews, sorted/filtered slices, or explicit 'first N rows' / 'top N' queries (capped at 500 rows per call). Do NOT use this to retrieve a full dataset, 'all rows', or anything the user wants to analyze in Excel / pandas / duckdb - use export_data for those. Data is only available after the workflow run has completed (status is no longer 'Running' or 'Validating'). Do NOT poll or sleep-wait for completion.",
+    description: "Get a PAGE of extracted data from a workflow. Use ONLY for previews, sorted/filtered slices, or explicit 'first N rows' / 'top N' queries (capped at 500 rows per call). Do NOT use this to retrieve a full dataset, 'all rows', or anything the user wants to analyze in Excel / pandas / duckdb - use export_data for those. Data is only available after the workflow run has completed (status is no longer 'Running' or 'Verifying data'). Do NOT poll or sleep-wait for completion.",
     inputSchema: strictSchema({
       workflowId: z12.string().min(1).describe("The workflow ID"),
       limit: limitInput({
@@ -3598,7 +3679,7 @@ function registerWorkflowRunTools(server, ctx, helpers) {
     try {
       const workflow = await ctx.client.workflow.get(args.workflowId);
       workflowName = workflow.name || args.workflowId;
-      workflowStatus = deriveStatusLabel(workflow);
+      workflowStatus = workflowStatusLabel(workflow);
     } catch {}
     if (!args.confirmed) {
       return jsonResult({
@@ -3710,6 +3791,7 @@ var init_workflows = __esm(() => {
   init_output_schemas();
   init_pagination();
   init_schemas();
+  init_workflow_status();
   WORKFLOW_AUDIT_WATCHED_KEYS = [
     "name",
     "tags",
@@ -3727,9 +3809,9 @@ var init_workflows = __esm(() => {
 function registerTools(server, ctx, capabilities) {
   const helpers = { withErrorHandling: createWithErrorHandling(ctx) };
   if (capabilities.scrape)
-    registerScrapeTools(server, ctx, helpers);
+    registerScrapeTools(server, ctx, helpers, capabilities.defaultProductType);
   registerWhoamiTool(server, ctx, helpers);
-  registerWorkflowCoreTools(server, ctx, helpers);
+  registerWorkflowCoreTools(server, ctx, helpers, capabilities.defaultProductType);
   registerAssistantTools(server, ctx, helpers);
   registerWorkflowHistoryTools(server, ctx, helpers);
   registerObservabilityTools(server, ctx, helpers);
@@ -4413,12 +4495,41 @@ var init_request_log = __esm(() => {
 
 // src/server.ts
 import { McpServer as McpServer2 } from "@modelcontextprotocol/server";
+function lifecycleInstruction2(defaultProductType) {
+  return defaultProductType === "realtime" ? "For one-time or scheduled workflows, the lifecycle is create_workflow → get_workflow (check status) → fetch_data (get results). Workflows run asynchronously - never poll or sleep-wait." : "Workflow lifecycle: create_workflow → get_workflow (check status) → fetch_data (get results). Workflows run asynchronously - never poll or sleep-wait.";
+}
+function routingInstructions2(defaultProductType) {
+  if (defaultProductType === "realtime") {
+    return [
+      `This team's default product is realtime monitoring. Use create_realtime_monitor whenever the user asks to monitor, track, or watch a page, or to get or extract data from it, unless they explicitly ask for a one-time run (e.g. "once") or a schedule (e.g. "daily"). A request that names no frequency is not explicit: default to realtime without asking the user to choose. It first persists notification channels, then asynchronously returns workflow/session/thread/job IDs; use the workflow Assistant tools for follow-up status, questions, and controls.`,
+      "Use create_workflow only when the user explicitly asks for a one-time run or a schedule. Use scrape for an immediate raw HTML or markdown fetch from one URL."
+    ];
+  }
+  return [
+    "Use create_realtime_monitor only when the user wants continuous realtime change detection with alerts. It first persists notification channels, then asynchronously returns workflow/session/thread/job IDs; use the workflow Assistant tools for follow-up status, questions, and controls.",
+    "For one-time or scheduled extraction, use create_workflow. Use scrape for an immediate raw HTML or markdown fetch from one URL. Use create_workflow for structured extraction, recurring runs, monitoring, or navigation-heavy jobs."
+  ];
+}
 async function createServer3(auth, options) {
   const ctx = {
     client: createKadoaClient(auth),
     principal: auth.principal,
     teamId: auth.principal.teamId
   };
+  let capabilities = options?.capabilities;
+  if (!capabilities) {
+    try {
+      const response = await ctx.client.user.getFeatures();
+      capabilities = {
+        scrape: response.features.scrape,
+        defaultProductType: parseDefaultProductType(response.defaultProductType)
+      };
+    } catch (error) {
+      const message = error instanceof Error ? error.message : "unknown error";
+      console.error(`[CAPABILITIES] Failed to resolve team capabilities; gated tools will be hidden: ${message}`);
+      capabilities = { scrape: false };
+    }
+  }
   const server = new McpServer2({ name: "kadoa", version: package_default.version }, {
     instructions: [
       "IMPORTANT: One workflow = one source. A single workflow can extract many different fields, tables, and sections from the same URL(s) using a rich schema.",
@@ -4427,15 +4538,14 @@ async function createServer3(auth, options) {
       "Kadoa workflows use agentic navigation: the AI agent can browse pages, click buttons, fill forms, select dropdowns, paginate, open detail pages, and handle multi-step interactions.",
       "Describe the full navigation steps in the prompt (e.g., 'select year 2022 from the dropdown, click Search, extract the table, then repeat for 2023').",
       "",
-      "Workflow lifecycle: create_workflow → get_workflow (check status) → fetch_data (get results). Workflows run asynchronously - never poll or sleep-wait.",
+      lifecycleInstruction2(capabilities.defaultProductType),
       "",
-      "Use create_realtime_monitor only when the user wants continuous realtime change detection with alerts. It first persists notification channels, then asynchronously returns workflow/session/thread/job IDs; use the workflow Assistant tools for follow-up status, questions, and controls.",
-      "For one-time or scheduled extraction, use create_workflow. Use scrape for an immediate raw HTML or markdown fetch from one URL. Use create_workflow for structured extraction, recurring runs, monitoring, or navigation-heavy jobs.",
+      ...routingInstructions2(capabilities.defaultProductType),
       "If a one-time or scheduled workflow fails, retry it with run_workflow using the existing workflow ID. Do NOT delete and recreate the workflow just to retry it - preserve its workflow ID and configuration. Realtime workflows cannot be manually run.",
       "Use list_changes and get_change to retrieve detected diffs from realtime monitoring workflows.",
       "List tools, plus fetch_data, get_workflow_history, and get_workflow_assistant_timeline, all return pagination {totalCount?, hasMore, next?}. When hasMore is true, call the same tool again with pagination.next as its arguments to get the next page. totalCount is present only when the total is known; never treat the page size as the total.",
       "Large results (fetch_data, get_workflow_history, get_change) are cut to whole rows or entries under about 80 KB and marked truncated: true with a hint; follow pagination.next, narrow the filters, or use export_data, and never conclude the data ends there.",
-      "When the user asks what needs their attention, call list_workflows with statusFilters group:attention. A row with awaitingUserInput.prompt is the Assistant's pending question: answer it with get_workflow_assistant then answer_workflow_assistant_question. A row with awaitingUserInput but no prompt is a setup the user left unfinished: use get_workflow_assistant to see where it stopped, then resume_workflow_assistant or request_workflow_update. A row with displayState PREVIEW and no awaitingUserInput is preview data to review or approve with approve_workflow. Workflows with an open support ticket are excluded from group:attention and appear under group:support.",
+      `When the user asks what needs their attention, call list_workflows with statusFilters group:attention. A row with awaitingUserInput.prompt is the Assistant's pending question: answer it with get_workflow_assistant then answer_workflow_assistant_question. A row with awaitingUserInput but no prompt is a setup the user left unfinished: use get_workflow_assistant to see where it stopped, then resume_workflow_assistant or request_workflow_update. A row with status "Preview ready" is preview data ready to approve with approve_workflow; "Preview" means preview data exists but approval isn't available yet. Workflows with an open support ticket are excluded from group:attention and appear under group:support.`,
       "",
       "Schema tips: Use descriptive field names and examples. Group related data under one entity.",
       "The AI agent uses the schema + prompt to understand what to extract - a detailed prompt with a comprehensive schema produces better results than multiple simple workflows.",
@@ -4444,17 +4554,6 @@ async function createServer3(auth, options) {
     ].join(`
 `)
   });
-  let capabilities = options?.capabilities;
-  if (!capabilities) {
-    try {
-      const response = await ctx.client.user.getFeatures();
-      capabilities = { scrape: response.features.scrape };
-    } catch (error) {
-      const message = error instanceof Error ? error.message : "unknown error";
-      console.error(`[CAPABILITIES] Failed to resolve team capabilities; gated tools will be hidden: ${message}`);
-      capabilities = { scrape: false };
-    }
-  }
   registerTools(server, ctx, capabilities);
   server.server.onerror = (error) => console.error("[MCP Error]", safeErrorSummary(error));
   return server;
@@ -4534,7 +4633,7 @@ async function requestNeedsCapabilityLookup(request) {
       return false;
     }
   }
-  return method === "tools/list" || method === "tools/call" && name === "scrape";
+  return method === "initialize" || method === "server/discover" || method === "tools/list" || method === "tools/call" && name === "scrape";
 }
 function isClientCausedHandlerError(error) {
   if (error instanceof ProtocolError)
@@ -4874,12 +4973,41 @@ init_package();
 init_client();
 init_tools();
 import { McpServer } from "@modelcontextprotocol/server";
+function lifecycleInstruction(defaultProductType) {
+  return defaultProductType === "realtime" ? "For one-time or scheduled workflows, the lifecycle is create_workflow → get_workflow (check status) → fetch_data (get results). Workflows run asynchronously - never poll or sleep-wait." : "Workflow lifecycle: create_workflow → get_workflow (check status) → fetch_data (get results). Workflows run asynchronously - never poll or sleep-wait.";
+}
+function routingInstructions(defaultProductType) {
+  if (defaultProductType === "realtime") {
+    return [
+      `This team's default product is realtime monitoring. Use create_realtime_monitor whenever the user asks to monitor, track, or watch a page, or to get or extract data from it, unless they explicitly ask for a one-time run (e.g. "once") or a schedule (e.g. "daily"). A request that names no frequency is not explicit: default to realtime without asking the user to choose. It first persists notification channels, then asynchronously returns workflow/session/thread/job IDs; use the workflow Assistant tools for follow-up status, questions, and controls.`,
+      "Use create_workflow only when the user explicitly asks for a one-time run or a schedule. Use scrape for an immediate raw HTML or markdown fetch from one URL."
+    ];
+  }
+  return [
+    "Use create_realtime_monitor only when the user wants continuous realtime change detection with alerts. It first persists notification channels, then asynchronously returns workflow/session/thread/job IDs; use the workflow Assistant tools for follow-up status, questions, and controls.",
+    "For one-time or scheduled extraction, use create_workflow. Use scrape for an immediate raw HTML or markdown fetch from one URL. Use create_workflow for structured extraction, recurring runs, monitoring, or navigation-heavy jobs."
+  ];
+}
 async function createServer(auth, options) {
   const ctx = {
     client: createKadoaClient(auth),
     principal: auth.principal,
     teamId: auth.principal.teamId
   };
+  let capabilities = options?.capabilities;
+  if (!capabilities) {
+    try {
+      const response = await ctx.client.user.getFeatures();
+      capabilities = {
+        scrape: response.features.scrape,
+        defaultProductType: parseDefaultProductType(response.defaultProductType)
+      };
+    } catch (error) {
+      const message = error instanceof Error ? error.message : "unknown error";
+      console.error(`[CAPABILITIES] Failed to resolve team capabilities; gated tools will be hidden: ${message}`);
+      capabilities = { scrape: false };
+    }
+  }
   const server = new McpServer({ name: "kadoa", version: package_default.version }, {
     instructions: [
       "IMPORTANT: One workflow = one source. A single workflow can extract many different fields, tables, and sections from the same URL(s) using a rich schema.",
@@ -4888,15 +5016,14 @@ async function createServer(auth, options) {
       "Kadoa workflows use agentic navigation: the AI agent can browse pages, click buttons, fill forms, select dropdowns, paginate, open detail pages, and handle multi-step interactions.",
       "Describe the full navigation steps in the prompt (e.g., 'select year 2022 from the dropdown, click Search, extract the table, then repeat for 2023').",
       "",
-      "Workflow lifecycle: create_workflow → get_workflow (check status) → fetch_data (get results). Workflows run asynchronously - never poll or sleep-wait.",
+      lifecycleInstruction(capabilities.defaultProductType),
       "",
-      "Use create_realtime_monitor only when the user wants continuous realtime change detection with alerts. It first persists notification channels, then asynchronously returns workflow/session/thread/job IDs; use the workflow Assistant tools for follow-up status, questions, and controls.",
-      "For one-time or scheduled extraction, use create_workflow. Use scrape for an immediate raw HTML or markdown fetch from one URL. Use create_workflow for structured extraction, recurring runs, monitoring, or navigation-heavy jobs.",
+      ...routingInstructions(capabilities.defaultProductType),
       "If a one-time or scheduled workflow fails, retry it with run_workflow using the existing workflow ID. Do NOT delete and recreate the workflow just to retry it - preserve its workflow ID and configuration. Realtime workflows cannot be manually run.",
       "Use list_changes and get_change to retrieve detected diffs from realtime monitoring workflows.",
       "List tools, plus fetch_data, get_workflow_history, and get_workflow_assistant_timeline, all return pagination {totalCount?, hasMore, next?}. When hasMore is true, call the same tool again with pagination.next as its arguments to get the next page. totalCount is present only when the total is known; never treat the page size as the total.",
       "Large results (fetch_data, get_workflow_history, get_change) are cut to whole rows or entries under about 80 KB and marked truncated: true with a hint; follow pagination.next, narrow the filters, or use export_data, and never conclude the data ends there.",
-      "When the user asks what needs their attention, call list_workflows with statusFilters group:attention. A row with awaitingUserInput.prompt is the Assistant's pending question: answer it with get_workflow_assistant then answer_workflow_assistant_question. A row with awaitingUserInput but no prompt is a setup the user left unfinished: use get_workflow_assistant to see where it stopped, then resume_workflow_assistant or request_workflow_update. A row with displayState PREVIEW and no awaitingUserInput is preview data to review or approve with approve_workflow. Workflows with an open support ticket are excluded from group:attention and appear under group:support.",
+      `When the user asks what needs their attention, call list_workflows with statusFilters group:attention. A row with awaitingUserInput.prompt is the Assistant's pending question: answer it with get_workflow_assistant then answer_workflow_assistant_question. A row with awaitingUserInput but no prompt is a setup the user left unfinished: use get_workflow_assistant to see where it stopped, then resume_workflow_assistant or request_workflow_update. A row with status "Preview ready" is preview data ready to approve with approve_workflow; "Preview" means preview data exists but approval isn't available yet. Workflows with an open support ticket are excluded from group:attention and appear under group:support.`,
       "",
       "Schema tips: Use descriptive field names and examples. Group related data under one entity.",
       "The AI agent uses the schema + prompt to understand what to extract - a detailed prompt with a comprehensive schema produces better results than multiple simple workflows.",
@@ -4905,17 +5032,6 @@ async function createServer(auth, options) {
     ].join(`
 `)
   });
-  let capabilities = options?.capabilities;
-  if (!capabilities) {
-    try {
-      const response = await ctx.client.user.getFeatures();
-      capabilities = { scrape: response.features.scrape };
-    } catch (error) {
-      const message = error instanceof Error ? error.message : "unknown error";
-      console.error(`[CAPABILITIES] Failed to resolve team capabilities; gated tools will be hidden: ${message}`);
-      capabilities = { scrape: false };
-    }
-  }
   registerTools(server, ctx, capabilities);
   server.server.onerror = (error) => console.error("[MCP Error]", safeErrorSummary(error));
   return server;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@kadoa/mcp",
-  "version": "0.7.0-rc.2",
+  "version": "0.7.1",
   "description": "Kadoa MCP Server — manage workflows from Claude Desktop, Cursor, and other MCP clients",
   "type": "module",
   "main": "dist/index.js",
@@ -27,7 +27,7 @@
     "prepublishOnly": "bun run check-types && bun run test:unit && bun run build"
   },
   "dependencies": {
-    "@kadoa/node-sdk": "^0.45.0",
+    "@kadoa/node-sdk": "^0.48.0",
     "@modelcontextprotocol/client": "2.1.0",
     "@modelcontextprotocol/express": "2.0.1",
     "@modelcontextprotocol/node": "2.1.0",