dataflows-consumption-cli

Inspect, monitor, and query saved Microsoft Fabric Dataflows Gen2 via read-only CLI operations.

Updated Jul 14, 2026
One-click install
npx skills add https://github.com/9vantage/skills-for-fabric-clone --skill dataflows-consumption-cli-9vantage
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: dataflows-consumption-cli
Source: https://github.com/9vantage/skills-for-fabric-clone/tree/main/plugins/fabric-skills/skills/dataflows-consumption-cli
Command: npx skills add https://github.com/9vantage/skills-for-fabric-clone --skill dataflows-consumption-cli-9vantage

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires pyarrow, pandas, and includes references (resource) components.

What problem does it solve? It gives AI agents a safe, read-only way to explore Microsoft Fabric Dataflows Gen2 — listing dataflows, decoding base64 definitions (mashup.pq, queryMetadata.json), discovering parameters, tracking refresh history, and executing saved queries — without any risk of mutating or deleting workspace items. ## Core Features & Use Cases - Definition Inspection: Decode base64 definition parts to read the Power Query M code, connection metadata, and platform config of any saved dataflow. - Refresh & Job Monitoring: Retrieve job instances, failure reasons, and refresh status with LRO polling and exponential backoff patterns. - Query Execution: Run persisted or ad-hoc read-only queries via the executeQuery endpoint, parse the Apache Arrow IPC response, render results as markdown tables, or convert them to CSV. - Use Case: Ask the agent to "show the refresh history and M code for the Sales Data Pipeline dataflow" and it will discover the workspace, decode the definition, and summarize recent job failures. ## Quick Start Ask the agent to list all dataflows in your Fabric workspace and show the refresh status of a specific one.

Frequently Asked Questions about dataflows-consumption-cli

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I list all dataflows in a Microsoft Fabric workspace?

Use az rest with a GET request to the /v1/workspaces/{workspaceId}/dataflows endpoint, passing --resource "https://api.fabric.microsoft.com" for the correct token audience. Filter results client-side with JMESPath since there is no server-side name filter.

How do I decode the Power Query M code inside a Fabric dataflow?

Call the getDefinition endpoint with a POST request (not GET), then base64-decode the mashup.pq part from the response. Large definitions may return 202 with a Location header, so poll the operation URL until it completes.

Can this skill create, update, or delete a dataflow?

No. The skill is strictly read-only and refuses any create, update, or delete operation, including updateDefinition and Delete Dataflow calls. For authoring or persisting changes, use the dataflows-authoring-cli skill instead.

Why does executeQuery return HTTP 200 even when my query fails?

The executeQuery endpoint returns 200 with an Arrow stream even when the underlying source query fails, embedding the error JSON inside the stream's PQ Arrow Metadata. Always grep the saved Arrow file for an "Error" marker before parsing it with pyarrow.

Why do I get a 401 Unauthorized error when calling the Fabric API?

A 401 usually means the token audience is wrong or the session expired. Run az login again and always pass --resource "https://api.fabric.microsoft.com" to az rest so the token is issued for the Fabric API.

What permissions are needed to read a dataflow definition?

The getDefinition operation requires Contributor role or higher (Read+Write) on the workspace; a Viewer role returns 403 Forbidden. Listing dataflows and reading job history work with lower-privilege read access.