Pbi deployer

Skill yujiyamane/pbi-ai-skills/.github/skills/pbi-deployer

The Power BI Skill Pack for Claude Code

Install

npx -y skills add yujiyamane/pbi-ai-skills --skill pbi-deployer

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Deploy a Power BI Project (PBIP) to Power BI Service / Microsoft Fabric using fabric-cicd. One-command deploy from a local PBIP folder to a target workspace. Use when user says: 'deploy dashboard', 'publish to Power BI', 'deploy to fabric', 'push PBIP', 'run deployer', 'deploy this report', 'fabric deploy', or provides a PBIP path with deployment intent.

SKILL.md

3.3 KB, as published. Nobody here has run it

PBI Deployer

Overview

One-command deploy of a Power BI Project (PBIP) to Power BI Service or Microsoft Fabric using fabric-cicd. Handles authentication, workspace resolution, and deployment in a single step.

Trigger phrases: deploy dashboard, publish to Power BI, deploy to fabric, push PBIP, run deployer, deploy this report, fabric deploy

Prerequisites

pip install fabric-cicd
az login

Verify both are ready:

fabric-cicd --version
az account show

If az login is needed, prompt the user to run it in their terminal (interactive auth cannot be automated).

Workflow

1. Collect inputs

Ask the user (in one message if not already provided):

PBIP path — full path to the .pbip file or project folder
Target workspace name — the Power BI / Fabric workspace to deploy to
Environment — dev, test, or prod (default: dev)

2. Validate PBIP structure

Confirm:

.pbip file exists
.Report/ subfolder exists
.SemanticModel/ subfolder exists

If any are missing, stop and report the issue.

3. Pre-deployment check

fabric-cicd validate --source "<pbip_folder>" --workspace "<workspace_name>"

Show the validation output. If validation fails, stop and show the error — do not proceed to deploy.

4. Deploy

fabric-cicd deploy \
  --source "<pbip_folder>" \
  --workspace "<workspace_name>" \
  --environment "<environment>"

Show the full output in real time. Do not suppress errors.

5. Verify deployment

After a successful deploy:

fabric-cicd status --workspace "<workspace_name>" --item "<report_name>"

Report the deployment status to the user.

6. Report to user

✅ Deployment complete

Dashboard: <report_name>
Workspace: <workspace_name>
Environment: <environment>
URL: https://app.powerbi.com/groups/<workspace_id>/reports/<report_id>

Next: Open in Power BI Service and trigger a dataset refresh.

Error Handling

Error	Action
`az login` required	Ask user to run `az login` in their terminal first
Workspace not found	Check workspace name spelling and user permissions
Validation failure	Show full `fabric-cicd validate` output
Deploy failure	Show full error traceback — never suppress
Timeout	Suggest re-running; fabric-cicd deployments are idempotent

Rules

Never skip validation — always run fabric-cicd validate before deploy.
Show full output — never suppress errors or summarise tracebacks.
Idempotent — re-running deploy on the same PBIP is safe; it updates the existing report in the workspace.
Environment matters — prod deployments should be confirmed by the user before proceeding (ask: "You're deploying to prod — confirm?").
Auth is interactive — az login must be run by the user in their terminal; Claude Code cannot perform interactive browser auth.