@agent-native/core-corpus 0.204.0 → 0.204.1-nightly-20261008034100

package/corpus/README.md

@@ -32,4 +32,4 @@ rg -n "defineAction|useActionQuery" node_modules/@agent-native/core-corpus/corpu
 
 ## Generated Counts
 
-- template files: 11386
+- template files: 11398

package/corpus/templates/analytics/.agents/skills/adhoc-analysis/SKILL.md

@@ -36,7 +36,7 @@ Analytics result or direct the user to an Extensions page.
 
 ### Step 1: Understand the Question (catalog-first, clarify-first)
 
-Orient before gathering data. Consult the injected `<data-dictionary>` and data-source status first to see which sources are configured and which one owns each fact, then settle scope:
+Orient before gathering data. Consult the preloaded `<resource scope="analytics-catalog">` references (else one `search-analytics-query-catalog`) and data-source status first to see which sources are configured and which one owns each fact, then settle scope:
 
 - What is being analyzed? (deals, users, campaigns, errors, etc.)
 - What time range?

package/corpus/templates/analytics/.agents/skills/cross-source-analysis/SKILL.md

@@ -34,7 +34,8 @@ lock and contract-metric rules before querying the warehouse.
 
 Orient before fanning out:
 
-1. Read the injected `<data-dictionary>` and check data-source status to see
+1. Read the preloaded `<resource scope="analytics-catalog">` references (else
+   one `search-analytics-query-catalog`) and check data-source status to see
    which sources are configured and what each one actually holds.
 2. Map each fact in the question to the one source that owns it. Write a tiny
    plan: "identities + emails from HubSpot, usage from BigQuery, first-party

package/corpus/templates/analytics/.agents/skills/dashboard-management/SKILL.md

@@ -1,7 +1,7 @@
 ---
 name: dashboard-management
 description: >-
-  Create, edit, organize, or share Analytics dashboards and panels: sources, layout, mutate-dashboard edits and verification. Use before any dashboard or panel change.
+  Create, lay out, organize, or share Analytics dashboards and panels: sources, placement, folders, verification. Use to create a dashboard or to move, reorder, or lay out panels; a small edit of one existing panel needs no skill.
 ---
 
 # Dashboard Management
@@ -117,9 +117,9 @@ an ordinary activity scan.
 
 When the user asks for a dashboard:
 
-1. Read the injected `<data-dictionary>` block first (catalog-first). If relevant entries exist, use their `table`, `columns`, `queryTemplate`, and gotchas verbatim.
+1. Start from the preloaded `<resource scope="analytics-catalog">` references (catalog-first), else one `search-analytics-query-catalog`. If a relevant entry exists, use its `table`, `columns`, `queryTemplate`, and gotchas verbatim.
 2. If a metric definition, date range, or grain is ambiguous and the choice would change the panel's numbers, use the `ask-question` clarifying tool once before building. Skip it when the dictionary or the user already settled it.
-3. If a metric is not documented, do not guess column names. Ask for the table/columns or introspect the provider schema, then propose a dictionary entry with `save-data-dictionary-entry`.
+3. If a metric is not documented, do not guess column names and do not ask the user for them. Find the table and columns with `search-bigquery-schema` (or the provider's own schema action), then propose a dictionary entry with `save-data-dictionary-entry`.
 4. Build a complete `SqlDashboardConfig` with `name` and `panels`. Optionally set top-level `columns` (1–6, default 2) to control how many grid columns the panels before any section use.
 5. Every panel needs `id`, `title`, `source`, `chartType`, `width`, and `sql`. `width` is the number of grid columns the panel spans (1..6, clamped to the active section's column count). Section panels skip `source` and `sql` and may set their own `columns` (1–6) to override the dashboard default for the panels following the section. Extension panels (`chartType: "extension"`) also skip `source` and `sql`; use `config.extensionId` for ordinary author-selected shared embeds. Use `config.extensionSlotId` only when the user explicitly asks for a personal/per-viewer slot (see "Embedding An Extension As A Panel").
 6. Persist with `update-dashboard` (load it with `tool-search` — it is not on the

package/corpus/templates/analytics/AGENTS.md

@@ -4,13 +4,13 @@ Analytics owns sources, queries, charts, and dashboards; dashboards are canonica
 
 ## Skills
 
-Search skills with `rg --hidden --follow`; read the exact linked guide before deeper work. App: `.agents/skills/data-querying/SKILL.md` (SQL, results, `/chart`), `.agents/skills/bigquery/SKILL.md`, `.agents/skills/hubspot/SKILL.md`, `.agents/skills/gong/SKILL.md`, `.agents/skills/prometheus/SKILL.md`, `.agents/skills/account-health/SKILL.md`, `.agents/skills/cross-source-analysis/SKILL.md`, `.agents/skills/dashboard-management/SKILL.md`, `.agents/skills/adhoc-analysis/SKILL.md`, `.agents/skills/analysis-workspace/SKILL.md`, `.agents/skills/provider-api/SKILL.md`, `.agents/skills/data-programs/SKILL.md`, `.agents/skills/creative-context/SKILL.md`, `.agents/skills/admin-surfaces/SKILL.md`, `.agents/skills/custom-blocks/SKILL.md` (extension panels), `.agents/skills/incident-investigation/SKILL.md`. Shared: `.agents/skills/actions/SKILL.md`, `.agents/skills/adding-a-feature/SKILL.md`, `.agents/skills/agent-native-docs/SKILL.md`, `.agents/skills/agent-native-toolkit/SKILL.md`, `.agents/skills/client-side-routing/SKILL.md`, `.agents/skills/context-awareness/SKILL.md`, `.agents/skills/customizing-agent-native/SKILL.md`, `.agents/skills/delegate-to-agent/SKILL.md`, `.agents/skills/external-agents/SKILL.md`, `.agents/skills/frontend-design/SKILL.md`, `.agents/skills/performance/SKILL.md`, `.agents/skills/portability/SKILL.md`, `.agents/skills/real-time-sync/SKILL.md`, `.agents/skills/reliable-mutations/SKILL.md`, `.agents/skills/secrets/SKILL.md`, `.agents/skills/security/SKILL.md`, `.agents/skills/self-modifying-code/SKILL.md`, `.agents/skills/shadcn-ui/SKILL.md`, `.agents/skills/sharing/SKILL.md`, `.agents/skills/storing-data/SKILL.md`, `.agents/skills/turn-into-skill/SKILL.md`, `.agents/skills/workspace-conventions/SKILL.md`.
+The skills summary lists every skill; find more with `rg --hidden --follow` in `.agents/skills/<name>/SKILL.md` and read one with `pnpm action docs-search --slug "skill-<name>"` before deeper work.
 
 Use local docs only (no web research): `pnpm action docs-search --query "<topic>"` and `pnpm action docs-search --slug "<slug>"`. Source examples: `pnpm action source-search --query "<pattern>"` or `pnpm action source-search --path <path>`.
 
 ## Data questions
 
-- Edit a panel on the open dashboard with `get-sql-dashboard` (`panelIds`), then `mutate-dashboard`. A dashboard edit is done only when `mutate-dashboard` returns `verified: true`. `noRenderAffected: true` means no chart changed. On `verified: false`, an error, or "the change isn't visible", follow `nextStep` and call `inspect-dashboard-panel` before saying anything; "Applied N ops", a raw `bigquery` result, or no warning banner is not proof.
+- Edit a panel on the open dashboard with `get-sql-dashboard` (`panelIds`), then `mutate-dashboard`; no skill read is needed (read `dashboard-management` to create a dashboard or move, reorder, or lay out panels). A dashboard edit is done only when `mutate-dashboard` returns `verified: true`. `noRenderAffected: true` means no chart changed. On `verified: false`, an error, or "the change isn't visible", follow `nextStep` and call `inspect-dashboard-panel` before saying anything; "Applied N ops", a raw `bigquery` result, or no warning banner is not proof.
 - Start from the closest query example (a preloaded reference, else one `search-analytics-query-catalog`); to build or clone a dashboard from another, call `search-dashboard-references`, then inspect matches with `get-sql-dashboard` or `get-explorer-dashboard` by `kind`. A match is context, not live data.
 - Use one bounded SQL or server-side `run-code` call for lists, filters, counts, or cohorts. If the catalog misses, make one discovery pass (`list-data-dictionary`, `search-bigquery-schema`, `data-source-status`) before querying; do not fan out per item or add unasked breakdowns.
 - Give a concise, evidence-backed answer with source, window, filters, sample size, join method, and caveats. Label figures “Unverified” if no live query ran; never cite the public `demo` source as real evidence unless asked.

package/corpus/templates/analytics/actions/bigquery.ts

@@ -5,6 +5,7 @@ import { track } from "@agent-native/core/tracking";
 import { z } from "zod";
 
 import { runQuery } from "../server/lib/bigquery";
+import { recoverFromSchemaMiss } from "../server/lib/bigquery-schema-recovery";
 
 function extractBigQueryMessage(message: string): string {
   const jsonStart = message.indexOf("{");
@@ -122,7 +123,7 @@ function stopForRepeatedBigQueryQuery(): never {
 
 export default defineAction({
   description:
-    "Query the user-configured BigQuery data warehouse. Use this when the user asks for warehouse SQL, BigQuery, or a data-dictionary metric/table that lives in BigQuery. If the user names a provider action such as Jira or Pylon, use that provider action first and do not use BigQuery unless the user explicitly asks for a warehouse copy. For a named customer or organization ID, resolve the canonical CRM/contract identity first and verify the returned rows carry the same customer and org/root-org identifiers. For account health, distinguish completed-month usage from current partial snapshots, contract metrics from similarly named platform metrics, total distinct contracted users from DAU/WAU, and actual usage from contracted capacity. Pass standard SQL via the `sql` arg. Do NOT use `db-query` for warehouse data (it only reaches the app's own SQL database). If a query fails with a schema or SQL error (unknown dataset/table/column, syntax), treat it as a normal debugging signal: inspect the real schema with `search-bigquery-schema` (or query INFORMATION_SCHEMA), correct the query based on the error, and run it again — a few corrective attempts are expected. Surface the error to the user only if it still fails after a few attempts or is non-recoverable (missing credentials, permission, quota). Never rerun identical failing SQL, and never substitute made-up numbers for data you could not query.",
+    "Query the user-configured BigQuery data warehouse. Use this when the user asks for warehouse SQL, BigQuery, or a data-dictionary metric/table that lives in BigQuery. If the user names a provider action such as Jira or Pylon, use that provider action first and do not use BigQuery unless the user explicitly asks for a warehouse copy. For a named customer or organization ID, resolve the canonical CRM/contract identity first and verify the returned rows carry the same customer and org/root-org identifiers. For account health, distinguish completed-month usage from current partial snapshots, contract metrics from similarly named platform metrics, total distinct contracted users from DAU/WAU, and actual usage from contracted capacity. Pass standard SQL via the `sql` arg. Do NOT use `db-query` for warehouse data (it only reaches the app's own SQL database). If a query fails with a schema or SQL error (unknown dataset/table/column, syntax), treat it as a normal debugging signal: use the `didYouMean` and `columns` the failure often carries, else inspect the real schema with `search-bigquery-schema` (or query INFORMATION_SCHEMA), correct the query based on the error, and run it again — a few corrective attempts are expected. Surface the error to the user only if it still fails after a few attempts or is non-recoverable (missing credentials, permission, quota). Never rerun identical failing SQL, and never substitute made-up numbers for data you could not query.",
   schema: z.object({
     sql: z.string().describe("SQL query to execute"),
   }),
@@ -172,11 +173,23 @@ export default defineAction({
         };
       }
       if (/BigQuery (API|poll) error/i.test(msg)) {
+        const message = extractBigQueryMessage(msg);
+        const recovery = await recoverFromSchemaMiss(
+          args.sql,
+          message,
+          context?.signal,
+        );
+        const foundSchema = Boolean(
+          recovery?.columns?.length || recovery?.didYouMeanTables?.length,
+        );
         return {
           error: "bigquery_query_failed",
-          message: extractBigQueryMessage(msg),
+          message,
           recoverable: true,
-          hint: "Likely a schema mismatch (wrong dataset, table, or column) or a SQL issue. Use search-bigquery-schema to get the exact datasets/tables/columns (or query INFORMATION_SCHEMA), correct the SQL based on this error, and run it again. Change the query based on the error — do not rerun identical SQL — and never substitute made-up numbers for data you could not query.",
+          hint: foundSchema
+            ? "The warehouse schema for this failure is below: correct the SQL using the exact names in `didYouMean`, `didYouMeanTables`, or `columns` and run it again. Do not rerun identical SQL, and never substitute made-up numbers for data you could not query."
+            : "Likely a schema mismatch (wrong dataset, table, or column) or a SQL issue. Use search-bigquery-schema to get the exact datasets/tables/columns (or query INFORMATION_SCHEMA), correct the SQL based on this error, and run it again. Change the query based on the error — do not rerun identical SQL — and never substitute made-up numbers for data you could not query.",
+          ...recovery,
         };
       }
       throw err;

package/corpus/templates/analytics/actions/mutate-dashboard.ts

@@ -202,8 +202,12 @@ function nonEmptyOperations(
 }
 
 const apiHelp =
-  "Compact script form of `operations`: JSON-literal calls on `dashboard` only (quote object keys; no variables, loops, or imports). Use it for short layout or config edits; the dashboard-management skill lists the methods. " +
-  `Examples: ${DASHBOARD_MUTATION_EXAMPLES[0]} ${DASHBOARD_MUTATION_EXAMPLES[2]}`;
+  "Compact script form of `operations`: JSON-literal calls on `dashboard` only (quote object keys; no variables, loops, or imports). Use it for short layout or config edits; call with only `returnTypes: true` to list every method. " +
+  `Examples: ${[0, 2, 3, 4, 7].map((index) => DASHBOARD_MUTATION_EXAMPLES[index]).join(" ")}`;
+
+const operationsHelp =
+  "Edits applied atomically in one save, by panel id. " +
+  'Examples: {"op":"updatePanel","panelId":"top-referrers","patch":{"title":"Top Referrers by Domain","width":2,"config":{"yFormatter":"percent"}}} {"op":"movePanels","panelIds":["dau","wau"],"position":"top"} {"op":"setFilterDefault","filterId":"emailFilter","value":"exclude_builder"}';
 
 const dryRunHelp = "Validate and verify without saving.";
 const allowEmptyResultHelp =
@@ -215,7 +219,7 @@ const agentInputSchema = z.object({
     .array(mutationOperationSchema)
     .max(MAX_DASHBOARD_MUTATION_OPERATIONS)
     .optional()
-    .describe("Edits applied atomically in one save, by panel id."),
+    .describe(operationsHelp),
   code: z
     .string()
     .max(MAX_DASHBOARD_MUTATION_CODE_LENGTH)
@@ -434,7 +438,7 @@ function helpResult() {
 export default defineAction({
   description:
     "Edit a SQL dashboard in ONE atomic save: move, insert, duplicate, remove, and edit panels by id (title, SQL, width, config), patch dashboard fields, or set filter defaults. Pass typed `operations`; `code` is a compact script form of the same edits. " +
-    "Place a panel in a visible row with nextToPanelId or rowNumber. First-party panel SQL must bind to a dashboard time filter (config.timeScope). Use `compose-dashboard` for catalog metrics and keep large SQL out of `code`. Read the existing panels with `get-sql-dashboard` first and match their chart types, widths, and config. The dashboard-management skill owns placement, time-scope, and config rules. " +
+    "Place a panel in a visible row with nextToPanelId or rowNumber. First-party panel SQL must bind to a dashboard time filter (config.timeScope). For catalog metrics load `compose-dashboard` with `tool-search`, and keep large SQL out of `code`. Read the existing panels with `get-sql-dashboard` first and match their chart types, widths, and config. Small edits of existing panels need no skill; the dashboard-management skill owns new-panel placement, time-scope, and config rules. " +
     "Before saving, agent calls run every changed panel the way the dashboard page does; a panel that would show 'No data', drop configured columns, or fail is refused with the reason and nothing is written. " +
     "The result's `verified` flag and per-panel `verification` (or `unverified` reasons) are the proof the edit renders: report only what they show, and on `verified:false`, an error, or a user report that nothing changed, call `inspect-dashboard-panel` before saying anything about the chart.",
   schema: z.object({
@@ -449,7 +453,7 @@ export default defineAction({
       .optional()
       .describe(apiHelp),
     operations: operationsInputSchema.describe(
-      "Edits applied atomically in one save, by panel id. Native callers pass an array; shell/legacy callers may pass a JSON string.",
+      `${operationsHelp} Native callers pass an array; shell/legacy callers may pass a JSON string.`,
     ),
     dryRun: z.boolean().optional().describe(dryRunHelp),
     allowEmptyResult: z.boolean().optional().describe(allowEmptyResultHelp),

package/corpus/templates/analytics/actions/search-bigquery-schema.ts

@@ -1,18 +1,17 @@
 import { defineAction } from "@agent-native/core/action";
 import { z } from "zod";
 
-import { getBigQueryProjectId } from "../server/lib/bigquery";
-import { getAccessToken } from "../server/lib/gcloud";
+import {
+  bigQueryGet,
+  flattenBigQueryFields,
+  getBigQueryProjectId,
+  getBigQueryTableMetadata,
+  listBigQueryTables,
+  type BigQueryTableMetadata,
+  type BigQueryTableSummary,
+} from "../server/lib/bigquery";
 import { cliBoolean } from "./schema-helpers";
 
-interface BigQueryField {
-  name: string;
-  type?: string;
-  mode?: string;
-  description?: string;
-  fields?: BigQueryField[];
-}
-
 interface DatasetListResponse {
   datasets?: Array<{
     datasetReference?: { projectId?: string; datasetId?: string };
@@ -22,45 +21,6 @@ interface DatasetListResponse {
   }>;
 }
 
-interface TableListResponse {
-  tables?: Array<{
-    tableReference?: {
-      projectId?: string;
-      datasetId?: string;
-      tableId?: string;
-    };
-    type?: string;
-    friendlyName?: string;
-    labels?: Record<string, string>;
-  }>;
-}
-
-interface BigQueryTableSummary {
-  projectId?: string;
-  datasetId?: string;
-  tableId?: string;
-  type?: string;
-  friendlyName?: string;
-  labels?: Record<string, string>;
-}
-
-interface TableMetadata {
-  tableReference?: {
-    projectId?: string;
-    datasetId?: string;
-    tableId?: string;
-  };
-  friendlyName?: string;
-  description?: string;
-  type?: string;
-  location?: string;
-  numRows?: string;
-  numBytes?: string;
-  timePartitioning?: unknown;
-  clustering?: unknown;
-  schema?: { fields?: BigQueryField[] };
-}
-
 const PROJECT_RE = /^[A-Za-z][A-Za-z0-9-]{4,61}[A-Za-z0-9]$/;
 const ID_RE = /^[A-Za-z0-9_]+$/;
 const GLOBAL_SEARCH_DATASET_LIMIT = 100;
@@ -116,67 +76,7 @@ function parseTableRef(
   );
 }
 
-async function bigQueryGet<T>(url: string): Promise<T> {
-  const token = await getAccessToken();
-  const res = await fetch(url, {
-    headers: {
-      Authorization: `Bearer ${token}`,
-      "Content-Type": "application/json",
-    },
-  });
-
-  if (!res.ok) {
-    const text = await res.text();
-    let detail = "";
-    try {
-      const parsed = JSON.parse(text) as { error?: { message?: string } };
-      detail = parsed.error?.message ?? "";
-    } catch {
-      // Fall through to include the raw response body below.
-    }
-    if (detail) {
-      throw new Error(detail);
-    }
-    throw new Error(
-      `BigQuery metadata request failed (${res.status}): ${text}`,
-    );
-  }
-
-  return (await res.json()) as T;
-}
-
-function flattenFields(
-  fields: BigQueryField[] | undefined,
-  prefix = "",
-): Array<{
-  name: string;
-  type?: string;
-  mode?: string;
-  description?: string;
-}> {
-  if (!fields?.length) return [];
-  const columns: Array<{
-    name: string;
-    type?: string;
-    mode?: string;
-    description?: string;
-  }> = [];
-
-  for (const field of fields) {
-    const name = prefix ? `${prefix}.${field.name}` : field.name;
-    columns.push({
-      name,
-      type: field.type,
-      mode: field.mode,
-      description: field.description,
-    });
-    columns.push(...flattenFields(field.fields, name));
-  }
-
-  return columns;
-}
-
-function compactTable(meta: TableMetadata, includeColumns: boolean) {
+function compactTable(meta: BigQueryTableMetadata, includeColumns: boolean) {
   const ref = meta.tableReference ?? {};
   return {
     projectId: ref.projectId,
@@ -190,11 +90,13 @@ function compactTable(meta: TableMetadata, includeColumns: boolean) {
     numBytes: meta.numBytes ? Number(meta.numBytes) : undefined,
     timePartitioning: meta.timePartitioning,
     clustering: meta.clustering,
-    columns: includeColumns ? flattenFields(meta.schema?.fields) : undefined,
+    columns: includeColumns
+      ? flattenBigQueryFields(meta.schema?.fields)
+      : undefined,
   };
 }
 
-function matchesSearch(meta: TableMetadata, search: string): boolean {
+function matchesSearch(meta: BigQueryTableMetadata, search: string): boolean {
   const terms = search
     .toLowerCase()
     .replace(/[._-]+/g, " ")
@@ -208,7 +110,7 @@ function matchesSearch(meta: TableMetadata, search: string): boolean {
     ref.tableId,
     meta.friendlyName,
     meta.description,
-    ...flattenFields(meta.schema?.fields).flatMap((column) => [
+    ...flattenBigQueryFields(meta.schema?.fields).flatMap((column) => [
       column.name,
       column.type,
       column.description,
@@ -247,22 +149,6 @@ async function listDatasets(projectId: string, limit: number, search: string) {
     .slice(0, limit);
 }
 
-async function listTables(projectId: string, datasetId: string, limit: number) {
-  const url = new URL(
-    `https://bigquery.googleapis.com/bigquery/v2/projects/${projectId}/datasets/${datasetId}/tables`,
-  );
-  url.searchParams.set("maxResults", String(Math.min(limit, 1000)));
-  const result = await bigQueryGet<TableListResponse>(url.toString());
-  return (result.tables ?? []).map((table) => ({
-    projectId: table.tableReference?.projectId,
-    datasetId: table.tableReference?.datasetId,
-    tableId: table.tableReference?.tableId,
-    type: table.type,
-    friendlyName: table.friendlyName,
-    labels: table.labels,
-  }));
-}
-
 async function searchAcrossDatasets(
   projectId: string,
   search: string,
@@ -291,7 +177,7 @@ async function searchAcrossDatasets(
 
     datasetsScanned += 1;
     const remaining = GLOBAL_SEARCH_TABLE_LIMIT - tables.length;
-    const listed = await listTables(
+    const listed = await listBigQueryTables(
       projectId,
       datasetId,
       Math.min(remaining, GLOBAL_SEARCH_TABLE_LIMIT),
@@ -321,11 +207,11 @@ async function searchAcrossDatasets(
     const settled = await Promise.allSettled(
       batch.map(async (table) => {
         if (!table.datasetId || !table.tableId) return null;
-        const metadata = await getTableMetadata(
+        const metadata = await getBigQueryTableMetadata({
           projectId,
-          table.datasetId,
-          table.tableId,
-        );
+          datasetId: table.datasetId,
+          tableId: table.tableId,
+        });
         return matchesSearch(metadata, search)
           ? compactTable(metadata, true)
           : null;
@@ -366,15 +252,6 @@ async function searchAcrossDatasets(
   };
 }
 
-async function getTableMetadata(
-  projectId: string,
-  datasetId: string,
-  tableId: string,
-) {
-  const url = `https://bigquery.googleapis.com/bigquery/v2/projects/${projectId}/datasets/${datasetId}/tables/${tableId}`;
-  return await bigQueryGet<TableMetadata>(url);
-}
-
 export default defineAction({
   description:
     "Search or describe BigQuery metadata for the configured warehouse. Use before writing SQL when the data dictionary does not already name the dataset, table, and columns. With no args, lists datasets. With search and no dataset, searches accessible tables and columns across the configured project so the user does not need to provide internal table names. With dataset, lists tables. With dataset + table or a dataset.table value, returns columns for that table.",
@@ -417,11 +294,7 @@ export default defineAction({
 
     if (args.table) {
       const ref = parseTableRef(configuredProjectId, args.dataset, args.table);
-      const meta = await getTableMetadata(
-        ref.projectId,
-        ref.datasetId,
-        ref.tableId,
-      );
+      const meta = await getBigQueryTableMetadata(ref);
       return {
         mode: "table",
         table: compactTable(meta, true),
@@ -442,7 +315,11 @@ export default defineAction({
     }
 
     const datasetId = assertIdentifier("dataset", args.dataset);
-    const tables = await listTables(configuredProjectId, datasetId, limit);
+    const tables = await listBigQueryTables(
+      configuredProjectId,
+      datasetId,
+      limit,
+    );
     const includeColumns = args.includeColumns === true || !!search;
 
     if (!includeColumns) {
@@ -469,7 +346,11 @@ export default defineAction({
     const metadata = await Promise.all(
       tables.slice(0, Math.min(tables.length, limit)).map((table) => {
         const tableId = table.tableId ?? "";
-        return getTableMetadata(configuredProjectId, datasetId, tableId);
+        return getBigQueryTableMetadata({
+          projectId: configuredProjectId,
+          datasetId,
+          tableId,
+        });
       }),
     );
 

package/corpus/templates/analytics/changelog/2026-10-07-analytics-answers-now-acknowledge-when-reference-catalogs-ar.md

@@ -0,0 +1,6 @@
+---
+type: fixed
+date: 2026-10-07
+---
+
+Analytics now treats chart filter changes as data requests and reports when reference catalogs are incomplete.

package/corpus/templates/analytics/changelog/2026-10-07-analytics-chat-follows-short-follow-ups-like-what-about-emea.md

@@ -0,0 +1,6 @@
+---
+type: improved
+date: 2026-10-07
+---
+
+Analytics chat follows short follow-ups like "what about EMEA?" and recovers from an unknown BigQuery column in one step

package/corpus/templates/analytics/changelog/2026-10-07-analytics-chat-no-longer-stays-on-running-after-the-agent-finishes.md

@@ -0,0 +1,6 @@
+---
+type: fixed
+date: 2026-10-07
+---
+
+Analytics chat no longer stays on "running" after the agent has finished; the thread reloads so the change is visible without asking again

package/corpus/templates/analytics/changelog/2026-10-07-analytics-chat-stops-retrying-an-unconnected-source.md

@@ -0,0 +1,6 @@
+---
+type: fixed
+date: 2026-10-07
+---
+
+Analytics chat stops retrying a data source that is not connected and says who can connect it

package/corpus/templates/analytics/changelog/2026-10-07-analytics-thumbs-down-has-reason-chips-and-copy-details.md

@@ -0,0 +1,6 @@
+---
+type: improved
+date: 2026-10-07
+---
+
+Thumbs-down on a chat answer now offers quick reasons and a Copy details button that copies the run id and thread link for a report

package/corpus/templates/analytics/server/lib/agent-chat-plan-mode.ts

@@ -2,6 +2,7 @@ export const INITIAL_TOOL_NAMES = [
   "describe-workspace-apps",
   "call-agent",
   "view-screen",
+  "ask-question",
   "data-source-status",
   "list-analyses",
   "get-analysis",