cube-build-content

Build Cube content

Writes state. Reports and dashboards are what people see, so a mistake here is visible to more than the person who made it.

Choose the path

Cube MCP tools are written cube:<tool> below: the cube server this plugin bundles. Through the Cube connector, the same tools carry the connector's name.

If the Cube MCP tools are available (this plugin's cube server or the Cube connector), use them for workbooks, reports, dashboards and folders:

1. Verify the query with cube:runQuery first — see cube-run-query.
2. cube:createWorkbook — returns workbookId and a url. Skip it when the user named a workbook to build into.
3. cube:createReport with that workbookId, one per chart. Reports take Cube SQL (sqlQuery) and a chartCategory. Omit workbookId only for a standalone exploration, which is what the Sheets and Excel add-ons list.
4. cube:readWorkbook, then cube:updateDashboard to lay the reports out.
5. cube:publishDashboard — returns the live url. Give it to the user.

Change an existing report in place with cube:readReport and cube:updateReport rather than recreating it. cube:manageFolders lists and creates folders.

Scheduled notifications, duplicating a workbook, and connecting a report to an external spreadsheet are CLI-only. Use the CLI path below for those, and for everything when the MCP tools are not connected. The order of operations and conventions apply to both paths.

Preflight (CLI)

command -v cube >/dev/null || echo "Cube CLI not installed: curl -fsSL https://raw.githubusercontent.com/cube-js/cube/master/install-cli.sh | sh"
cube whoami || echo "Not authenticated. Interactive: cube login. Headless: set CUBE_API_URL + CUBE_API_KEY."
cube context list

Order of operations

Content hangs off a workbook, so build outward from one:

1. Verify the query first. Run it with cube-run-query and confirm the numbers are right. Saving a wrong query is how a bad number gets an audience.
2. Create or pick a workbook.
3. Add reports to it.
4. Publish it as a dashboard if people need to consume it.
5. Schedule it if it should arrive on a cadence.

Workbooks

cube workbooks create <deployment> --name "Revenue review"
cube workbooks update <deployment> <workbook> --name "Revenue review Q3"
cube workbooks duplicate <deployment> <workbook>
cube workbooks dashboard <deployment> <workbook> --data '<dashboard-input>'
cube workbooks publish <deployment> <workbook>      # makes the dashboard
cube workbooks delete <deployment> <workbook>

Duplicating an existing workbook is usually a better starting point than creating an empty one — it inherits layout and conventions the team already uses.

Reports

cube reports create <deployment> --workbook <workbook> --name "Revenue by month" --json-query '<query>'
cube reports update <deployment> <report> --json-query '<query>'
cube reports refresh <deployment> <report>
cube reports delete <deployment> <report>

--json-query takes the same query shape cube-run-query sends to the load endpoint — measures, dimensions, timeDimensions, filters. Take the query you already verified rather than retyping it; retyping is where a filter goes missing.

Complex bodies go through -d/--data, which accepts inline JSON, @file.json or - for stdin. Dedicated flags override values in --data.

cube reports connect-workbook is for placing a report in an external spreadsheet, not attaching it to a Cube workbook. For that workflow use:

cube reports connect-workbook <deployment> <report> --external-workbook-id <id> --result-location <range>

Folders

cube folders create <deployment> --name "Finance"
cube folders update <deployment> <folder> --name "Finance & RevOps"
cube workspace move <deployment> --type WORKBOOK|REPORT|FOLDER --id <id> --folder <folder>   # omit --folder for the root
cube folders delete <deployment> <folder>

Scheduled notifications

cube notifications create <deployment> --data '{"scheduleType":"WEEKLY","dashboardPublicId":"<id>","dayOfWeek":1,"hour":9,"timezone":"Europe/London"}'
cube notifications recipients add <deployment> <notification> --data '{"recipients":[{"type":"USER","userId":<id>}]}'
cube notifications recipients list <deployment> <notification>
cube notifications update <deployment> <notification> --data '<UpdateNotificationInput JSON>'

scheduleType is HOURLY, DAILY, WEEKLY, MONTHLY or CUSTOM (with customCron), dayOfWeek runs 0–6 from Sunday, and hour is 0–23 in timezone. A recipient's type is USER, EMBED_USER or SLACK. For the full body, cube spec notification --json prints the schemas.

Before changing or deleting a schedule, list its recipients and say who is currently receiving it. People notice when a report stops arriving, and they rarely connect it to a change someone made.

Conventions

• Name things the way the user said them. Do not tidy "MRR" into "Monthly Recurring Revenue" unless asked.
• Report the id of everything you create, and the URL if the output carries one. The user's next step is opening it.
• Creating is cheap and deleting is not. When a request is ambiguous between "change the existing one" and "make a new one", make a new one and say so.

When something fails

Symptom	Cause
Report saves but shows no data	The query is wrong, not the report — verify with cube-run-query
Member not found	Stale member name; the model changed. Re-check with cube-explore-model
Publish does nothing visible	The workbook had no reports yet — add them first
403	The account can read the workspace but not write to it