Skip to content

zebrabi/zebra-bi-agentic / plugins/zebra-bi/skills/report-publishing

report-publishing

Verified

The Zebra-specific things to do around publishing a report to a Power BI or Fabric workspace. Microsoft's powerbi-report-management skill does the upload; this covers the visual-track decision that has to be made before you publish, the Zebra licence in the service, and the post-publish refresh without which a report built from a portable .pbip is silently empty. Use when the user says "publish", "upload", "push to Fabric", "deploy the report", "share it with the team", or asks why a published report shows no data. For authoring, use report-authoring.

FileKindSize
SKILL.mdskill md4.5 KB
references/unreachable-source.mdmarkdown2.8 KB
---
name: report-publishing
description: >-
  The Zebra-specific things to do around publishing a report to a Power BI or Fabric workspace.
  Microsoft's powerbi-report-management skill does the upload; this covers the visual-track
  decision that has to be made before you publish, the Zebra licence in the service, and the
  post-publish refresh without which a report built from a portable .pbip is silently empty.
  Use when the user says "publish", "upload", "push to Fabric", "deploy the report", "share it
  with the team", or asks why a published report shows no data. For authoring, use
  report-authoring.
---

# Publishing

**Microsoft publishes Power BI reports. Use their skill.** `powerbi-report-management` owns the
upload, download, update and delete, and it is good. It ships in the `powerbi-authoring` plugin of
`/plugin marketplace add microsoft/skills-for-fabric`. Do not reimplement any of it.

**If that plugin cannot be installed** (an organisation that allows only the official marketplace),
the `.pbix` Imports call in `references/unreachable-source.md` needs no plugin and has been run end
to end with data.

Three things Microsoft's skill cannot know about. That is the whole of this skill.

## 1. Before: pick the visual track, because publishing is when it stops being free

The authoring default is **Tables+ and Charts+**. They are Beta and **not Microsoft-certified**.
Tables, Charts and Cards on the certified track are.

**Both tracks publish and render in the service** — checked as a pair, same report and data, only
the two GUIDs different. Do not talk anyone off the default on a worry that Plus will not survive
the upload.

What Plus costs is everything *downstream*: **`exportToFile`, PDF export, PowerPoint export and
e-mail subscriptions** all refuse uncertified visuals and draw an error symbol. So if any of those
matter, swap `ZebraBITablesPlus…` → `ZebraBITables…` and `ZebraBIChartsPlus…` → `waterfall…` before
you publish. One-line change, query untouched, Cards unchanged.

Ask if you cannot tell: a report that publishes fine and then cannot be e-mailed to the board is
re-done late.

## 2. Before: do not promise a render you have not seen

Say what you actually checked; the claim table at the end is the vocabulary. A clean render on your
machine is not evidence about anyone else's.

## 3. After: refresh, or ship an empty report

**A model with a calculated-table partition can arrive unprocessed, and the report then renders
empty.** The shape to watch, and where this was seen, is a portable `.pbip` with its rows in
`DATATABLE(...)`. Every step reports success and the first query says:

```
"The query referenced calculated table 'Spend' which does not hold any data
 because it needs to be recalculated or refreshed."
```

**Do not go down the entity-name-mismatch trail**, which is where the generic advice points. The
bindings are already correct, so a TMDL diff comes back clean and costs an hour.

**And no refresh helps when the service cannot reach the source** — a personal workbook, a local
path, a gateway-less database. Publish the `.pbix`: `references/unreachable-source.md`.

One call, about five seconds:

```
POST https://api.powerbi.com/v1.0/myorg/groups/{ws}/datasets/{semanticModelId}/refreshes
     {"type":"Full","commitMode":"Transactional"}
```

Then check `value[0].status == "Completed"` on `GET .../refreshes`, and prove the data with one
query against the **published** model, compared to the source file rather than to another visual:

```
POST .../datasets/{id}/executeQueries
     {"queries":[{"query":"EVALUATE ROW(\"Rows\", COUNTROWS(T), \"Total\", SUM(T[Col]))"}]}
```

## What you may claim afterwards

| You ran | You may say |
|---|---|
| Upload returned 202 → Succeeded | The items were created. Nothing more |
| `getDefinition` round-trip matches | The service kept the files, Zebra GUIDs included |
| DAX against the published model matches source | The data is right |
| A human opened the report | It renders |
| `exportToFile` came back clean | It renders — **only on the certified track**; on Plus this check reports a failure that is not real |

Hand over the URL and say which line you reached:
`https://app.powerbi.com/groups/{workspaceId}/reports/{reportId}`

## If the Azure CLI is missing

`az rest` needs an admin install. The **Fabric CLI** is the no-admin substitute —
`pip install ms-fabric-cli` — and `fab api` maps one-for-one: `-X` method, `-i` body, `-q` JMESPath,
`--show_headers` for the LRO id, `-A powerbi` for a Power BI endpoint. Give it a path, not a full
URL, or every call 404s.
writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deploymentStars: 322kUpdated Claude CodeCodexVerifiedInstall
xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy.Stars: 119kUpdated Claude CodeCodexVerifiedInstall
childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release.Stars: 118kUpdated Claude CodeCodexVerifiedInstall