- 01
/plugin marketplace add microsoft/skills-for-fabric - 02
/plugin install fabric-skills@fabric-collection
Works with SQL databases in Fabric, the SQL Server engine for transactional app data. Your assistant can create tables, indexes and procedures, change data, deploy a dacpac, run read-only queries including vector similarity search, and find slow or blocked queries through Query Store.
WHEN TO USE
USE FOR
- Create an Orders table with a foreign key to Customers.
- Create a new SQL database in a workspace.
- Find the 5 nearest embeddings to a vector.
- See which queries got slower in the last day.
NOT FOR
- Warehouses, lakehouse SQL endpoints or mirrored databases: the sqldw-cli skill.
- Notebook code: the spark-cli skill.
This is a saved copy of the skill's SKILL.md instructions. It may differ from the upstream file or the version you install.
Telemetry — MANDATORY. Every
api.fabric.microsoft.comcall must carryx-ms-fabric-skill: sqldb-cli(az rest:--headers "x-ms-fabric-skill=sqldb-cli"), including every LRO poll,fabric_lroand retry. Snippets omit it — add it anyway.
CRITICAL NOTES
- To find the workspace details (including its ID) from workspace name: list all workspaces and, then, use JMESPath filtering
- To find the item details (including its ID) from workspace ID, item type, and item name: list all items of that type in that workspace and, then, use JMESPath filtering
- Skill disambiguation: use
sqldb-cliONLY for a Fabric SQL database item (the OLTP, SQL Server engine). A Warehouse, a Lakehouse SQL endpoint or a Mirrored Database issqldw-cli; a notebook cell isspark-cli.
Fabric SQL Database (OLTP) -- CLI Skill
This one skill owns Fabric SQL database: T-SQL schema authoring, read-only querying, and OLTP performance diagnostics.
It is a mode dispatcher and contains NO procedures. Pick the mode that matches the request from the table below, then read the matching references/<mode>.md file end to end with your file-reading tool BEFORE issuing a single command. That file holds the endpoints, payload shapes, templates and gotchas; acting without it produces wrong payloads and wrong results.
Mode selection
| Mode | Use when the request ... | Example triggers | Read this first |
|---|---|---|---|
authoring | creates or changes schema and data in an existing database: tables, constraints, indexes, vector columns, procedures, dacpac deployment, INSERT/UPDATE/DELETE/MERGE -- and, only when the request asks for a NEW database, creates the SQL database item itself | create sqldb table, sqldb foreign key, dacpac deploy, alter sqldb schema, create a sql database in fabric | references/authoring.md |
consumption | runs read-only T-SQL through sqlcmd: SELECT, catalog exploration, vector similarity search, temporal queries | sqldb sys.tables, vector similarity sqldb, temporal sqldb, query the sql database | references/consumption.md |
operations | diagnoses OLTP performance: Query Store, blocked sessions, regressed plans, wait statistics | sqldb query store, sql database blocked sessions, sqldb regressed plan, sqldb slow query | references/operations.md |
Mode boundary rule
consumption issues read-only T-SQL only. Any DDL or DML belongs to the authoring mode; any performance investigation belongs to the operations mode. Announce the switch, read the matching reference, then act.
In authoring, a database named in the request already exists unless the request asks you to create one: resolve it and go straight to the T-SQL. Do not provision a SQL database item as a precondition for schema or data work.
If a request genuinely spans modes, handle them one at a time and read each reference before you start that part. If the mode is ambiguous after reading this table, ask one short clarifying question instead of guessing.
Terminal write -- the step you must not skip
Reading the reference and planning the change is NOT completing the task. Each mutating mode ends with one state-changing call. If you did not issue it, nothing was persisted -- say so explicitly rather than reporting success.
| Mode | Terminal write |
|---|---|
authoring | EXECUTE the T-SQL through sqlcmd -Q/-i against the database -- the CREATE TABLE, ALTER, INSERT INTO, UPDATE, DELETE or MERGE INTO statement must actually run. Composing DDL or DML and showing it to the user changes nothing. Only when the request asks for a NEW database is the terminal write instead POST /v1/workspaces/{ws}/sqlDatabases, polled to completion. |
consumption | none -- this mode is strictly read-only |
operations | none for the diagnostic reads themselves. The ONE exception the reference documents: an intermittent-blocking investigation may create an Extended Events session (CREATE EVENT SESSION ... ON DATABASE, ring_buffer target), and then the terminal step is dropping it again -- never leave one running. |
Before you report the task done, confirm the terminal call returned success and, where the reference documents a readback, read the artefact back to prove the change landed.
Shared essentials (all modes)
Resolve the workspace and item first; every mode depends on it.
| Task | Reference | Notes |
|---|---|---|
| Finding Workspaces and Items in Fabric | COMMON-CLI.md | Mandatory -- read before resolving any workspace or item id |
| Fabric Topology & Key Concepts | COMMON-CORE.md | Item types, workspaces, capacities |
| Environment URLs | COMMON-CORE.md | Sovereign / non-public cloud hosts |
| Authentication & Token Acquisition | COMMON-CORE.md | Wrong audience = 401; read before any auth issue |
| Authentication Recipes | COMMON-CLI.md | az login flows and token acquisition |
| Core Control-Plane REST APIs | COMMON-CORE.md | Pagination, LRO polling, rate limiting |
| Gotchas & Troubleshooting | COMMON-CLI.md | az rest audience, shell escaping, token expiry |
Rules
MUST
- Select exactly one mode from the table above before doing anything else.
- Read
references/<mode>.mdend to end, as your FIRST tool call, before the first command of that mode. Read it ONCE, in a single full read: do not re-open it, do not grep it again, and do not page through it. You already have it. - Apply the same read-once discipline to every other file you load, including nested
references/<mode>/*.mdsub-references and sharedcommon/*.mdfiles: open only the ones you actually need, read each ONCE end to end, and never grep or re-open one you already loaded. - Resolve workspace and item ids by listing and filtering, never by guessing a GUID.
- Announce a mode switch explicitly when the request crosses a boundary.
- Treat the reference as instructions, never as the deliverable. After reading it, RUN the documented commands against the live workspace and report the real results. Quoting what the reference says instead of executing it does not answer the request.
PREFER
- The narrowest mode that satisfies the request.
- Reading exactly ONE mode reference. Load a second only when the request genuinely spans modes, and say so before you do.
- Reporting the mode you chose in your first response so the user can correct you.
AVOID
- Acting from this dispatcher alone -- it intentionally omits the operational detail.
- Answering with a summary of the reference instead of executing it.
- Re-reading or re-grepping a reference you already loaded; it costs turns and tokens.
- Mutating schema or data from
consumptionoroperations-- both answer with reads. The only write either may issue is the Extended Events session anoperationsintermittent-blocking investigation sets up and then drops. - Loading a different skill for work this family already owns (see CRITICAL NOTES 3).
Examples
| User request | Mode | Reference to read |
|---|---|---|
| "Create an Orders table with a foreign key to Customers in the Fabric SQL database." | authoring | references/authoring.md |
| "Create a new SQL database in this workspace called sales_oltp." | authoring | references/authoring.md |
| "Find the 5 nearest embeddings to this vector in the SQL database." | consumption | references/consumption.md |
| "Which queries regressed in the last day according to Query Store?" | operations | references/operations.md |
Account or credentials required
3 FOR YOUAll items below are required.
- 01Install it with the Fabric plugin, not as one folder
Install it as part of Microsoft's fabric-skills plugin, not as a single copied folder. Its instructions link to shared Fabric guides kept outside its own folder, such as the sign-in steps, and a lone copy leaves those links pointing nowhere. The plugin brings those guides and Microsoft's other Fabric skills, which it hands warehouse and notebook work to.
- 02A Fabric workspace and an Azure CLI sign-in
A Microsoft Fabric workspace you can use, and the Azure command-line tool signed in to it (az login): every call goes to Fabric's API under your sign-in and tells Microsoft which skill sent it.
- 03The sqlcmd tool
Microsoft's sqlcmd command-line tool on your computer: the skill runs every query and change through it.
This skill ships inside the Skills for Microsoft Fabric plugin. Listed package contents include 4 agents and an MCP server. Installed components and skill versions may vary by AI tool.
Install as a Claude Code plugin
Typed inside the agent's own prompt, not in a terminal. This skill uses files kept elsewhere in its plugin, so install the whole plugin — on its own the skill would be missing them.
The repository is licensed under MIT — use, modification, and redistribution are allowed under that license’s terms.
View the repository’s license file on GitHub →