docs: document undocumented customer-facing changes from the last week - #11652
docs: document undocumented customer-facing changes from the last week#11652keydunov wants to merge 1 commit into
Conversation
* docs(mcp): document runQuery's branchName param runQuery can now target a dev branch instead of the deployed model, closing the loop so an agent can verify a data-model edit before committing it (CUB-3732). * docs(chat): document saving a chat result as a standalone exploration Asking the agent to save a result without naming a workbook now saves it as a standalone exploration, matching what the Explore page's own Save does (CUB-4109). * docs(sheets, excel): document measure reordering and measure position The Measures pane in the pivot builder is reorderable, and a Measure position option controls whether measures nest under each column value or each measure spans all column values (CUB-2405). * docs(sheets, excel): document the add-in/add-on menu's account info The menu now shows the current tenant, Cube account, deployment, and signed-in user (CUB-4096). Co-authored-by: Claude <noreply@anthropic.com>
|
Preview deployment for your docs. Learn more about Mintlify Previews.
💡 Tip: Enable Workflows to automatically generate PRs for you. |
|
Claude finished @keydunov's task in 1m 50s —— View job Docs-only PR, no blockers — 0 high, 2 medium, 3 low. All 5 posted as inline comments.Full review
Verified
I could not verify the described behaviors against source — these features live in Findings
Note on the review checklistNothing here is code, so tests and linting are not applicable — the unchecked boxes in the PR description are fine as-is. Worth a local Aside (pre-existing, not from this PR)
|
| `runQuery` also takes an optional `branchName` to query a [dev branch][ref-dev-mode] | ||
| instead of the deployed model — use it to verify a data model edit before committing it. | ||
| Every response reports the branch it actually ran against (`null` means the deployed | ||
| model). Keep `branchName` identical across paginated calls: dropping it on a later page | ||
| re-runs the query against the deployed model instead of continuing on the branch. |
There was a problem hiding this comment.
A reader hitting this paragraph has no idea where a branchName value comes from — startDataModelEdit (the tool that returns one) isn't introduced until line 290, in a later section that's only registered for model-editing roles. Two small fixes:
- Point at the source of the value, e.g. "…use the
branchNamereturned bystartDataModelEdit, or the name of an existing dev branch." - The
runQuerytable row above describes the tool's other inputs/outputs (offset,hasMore,totalRows) but notbranchName, so someone skimming only the table misses it entirely. Worth a clause there too.
| model). Keep `branchName` identical across paginated calls: dropping it on a later page | ||
| re-runs the query against the deployed model instead of continuing on the branch. |
There was a problem hiding this comment.
"paginated calls" is ambiguous given the tool list above documents two pagination paths: runQuery itself paginates "via offset", while loadQueryResults "paginates through the results of a previous query." Which one does this caveat apply to?
If it's repeated runQuery calls with offset, say so ("when you re-issue runQuery with a higher offset"). If loadQueryResults is the normal path, the more important question is whether it carries the branch context implicitly or needs branchName passed too — as written a reader can't tell, and the failure mode (silently querying the deployed model) is exactly the one this paragraph is warning about.
| The add-on menu shows the current tenant, Cube account, deployment, and signed-in | ||
| user. If you want to revoke the authentication, open the add-on menu and click | ||
| **Sign out**. |
There was a problem hiding this comment.
Nit: the new sentence about what the menu displays is grafted onto the sign-out paragraph, so two unrelated facts share one block. Splitting reads better and makes the display change findable:
| The add-on menu shows the current tenant, Cube account, deployment, and signed-in | |
| user. If you want to revoke the authentication, open the add-on menu and click | |
| **Sign out**. | |
| The add-on menu shows the current tenant, Cube account, deployment, and signed-in | |
| user. | |
| If you want to revoke the authentication, open the add-on menu and click | |
| **Sign out**. |
Same applies to microsoft-excel.mdx:72-74.
| - **Semantic model integration** – All queries run against your semantic model with proper access control and security, honoring the active [security context](/docs/explore-analyze/workbooks/querying-data#applying-a-security-context)—including an override applied by a developer or admin | ||
| - **Queued messages** – Send follow-up messages while the agent is still processing | ||
| - **Save to Workbook** – Export insights to [Workbooks](/docs/explore-analyze/workbooks) for further analysis | ||
| - **Save to Workbook or as an exploration** – Ask the agent to save a result, either into a [workbook](/docs/explore-analyze/workbooks) as a report or, if you don't name a workbook, as a standalone [exploration](/docs/explore-analyze/explore) for further analysis |
There was a problem hiding this comment.
The behavior is accurate but the trigger is implicit — "if you don't name a workbook" means the destination depends on what the user omits from a prompt, which is hard to discover from a feature bullet. Consider a concrete example prompt pair somewhere in the body of the page (this page already has a "Try prompts like" pattern below), e.g. "Save this to the Revenue workbook" vs. "Save this".
Also worth a follow-up cross-link: explore.mdx documents "Exploring from Analytics Chat" as clicking Explore, and its "Saving explorations" section lists only the manual Save paths — neither mentions that the agent can now produce a saved exploration directly. A reader landing on Explore first won't learn about this.
| remove members from a query. Drag members within **Measures** to reorder them | ||
| in the output. | ||
|
|
||
| When a dimension is on **Columns**, the **Measure position** option controls | ||
| how measures nest relative to the column values: **After columns** (default) | ||
| nests measures under each column value, while **Before columns** makes each | ||
| measure span all column values. | ||
|
|
||
| <Frame> | ||
| <img src="https://ucarecdn.com/670fda9d-358a-4345-9242-7888d8aaedd8/" /> |
There was a problem hiding this comment.
The Measure position explanation is clear, and mirroring it verbatim in google-sheets.mdx matches how these two pages are already kept in sync — good.
One gap: the <Frame> screenshot immediately below predates both the reorderable Measures pane and the Measure position control, so the prose now describes UI the image doesn't show. Not blocking, but worth queuing a screenshot refresh (both pages).
Check List
Description of Changes Made
Recurring audit of recent
cube-js/cubeandcubedevinc/cubejs-enterprisechanges againstdocs-mintlify, filtered per the customer-facing criteria. Four undocumented, shipped, customer-facing changes:runQuery's new optionalbranchNameparam — lets an MCP client verify a data-model edit against its dev branch before committing (CUB-3732).Everything else reviewed in this window was either already documented in its own PR, an internal/super-admin-only change, or a partial rollout (dbt workspace publish, CUB-3964) better tracked separately once it completes — filed as a Linear ticket instead.
🤖 Generated with Claude Code
https://claude.ai/code/session_014z7xqaGDhByEBLWMT4SCe7
Generated by Claude Code