fabric-dw-mcp-cli
The fabric-dw MCP server provides comprehensive administration and management capabilities for Microsoft Fabric Data Warehouses and SQL Analytics Endpoints, with built-in security controls for destructive operations.
Workspace Management: List workspaces, get details, and set default collations.
Warehouse & SQL Endpoint Management: List, create, rename, delete, and take ownership of warehouses and SQL Analytics Endpoints; refresh endpoint metadata; get permissions (requires Fabric Admin role).
SQL Execution & Query Analysis: Execute arbitrary SQL (DDL, DML, SELECT) and retrieve estimated execution plans (SHOWPLAN_XML).
Monitoring & Performance: List running queries and active connections; terminate sessions; view request/session history, frequent queries, and long-running queries.
Audit Management: Get, enable, and disable SQL auditing; manage action groups and log retention periods.
Data Versioning & Recovery: List, create, rename, and delete snapshots and restore points; restore a warehouse in-place to a restore point.
Schema & Object Management: Manage schemas, views, stored procedures, and user-defined functions (scalar UDFs, inline TVFs) β list, get, create, update, drop, and rename.
Table Management: List, read, create, delete, truncate, clone, rename, and count rows in tables; load data from remote URLs (CSV/Parquet via COPY INTO).
Statistics Management: List, inspect (with histograms), create, update, and delete table statistics.
Warehouse Settings: Get/set result-set caching and time-travel retention periods.
SQL Pools (Beta): List, create, update, delete, enable, and disable custom SQL pools; view SQL pool insight events.
Utilities: Clear internal caches and generate dbt-fabric project files (profiles.yml, dbt_project.yml, sources, requirements, .gitignore).
Generates dbt-fabric project file contents for a Fabric Data Warehouse, enabling data transformation workflows with dbt.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@fabric-dw-mcp-cli list all data warehouses in my Fabric workspace"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Python CLI and MCP server for Microsoft Fabric Data Warehouses and SQL Analytics Endpoints: administer, query, optimize, and secure them from your terminal or your AI agent.
Full documentation: fdw.debruyn.dev
π£ Just announced! Read the story behind fabric-dw in the announcement blog post.
Description
fabric-dw provides two interfaces for managing Microsoft Fabric Data Warehouses and SQL Analytics Endpoints:
CLI: a command-line tool for common DW administration tasks.
MCP server: a Model Context Protocol server that exposes DW operations as tools for AI assistants.
Authentication is configured via the FABRIC_AUTH environment variable. The default (FABRIC_AUTH=default) uses azure-identity DefaultAzureCredential, which walks environment variables, Workload/Managed Identity, Azure CLI, Azure Developer CLI, Azure PowerShell, and interactive browser in order. Any of these will satisfy it. See the Authentication docs for the full chain, all supported sources, and debugging tips.
Related MCP server: mcp-fabric-api
Installation
pip install fabric-dw
# or run without installing:
uvx fabric-dw --help
# or install persistently on PATH:
uv tool install fabric-dwAfter installation, the fdw command is a short alias for fabric-dw; both invoke the same entry point. See the Install docs for MCP server setup, upgrading, and prerelease builds.
Quick Start
CLI
The workspace is a global root option -w / --workspace placed before the command group. Set a default once with fdw config set workspace <NAME> and omit -w on every subsequent call. Workspace resolution order: (1) -w flag, (2) FABRIC_DW_DEFAULT_WORKSPACE env var, (3) configured default.
# Run without installing; install to get the fdw alias
uvx fabric-dw --help
# Set a default workspace once; all subsequent commands pick it up
fdw config set workspace SalesWS# -- Run and explain SQL --
# Execute a query against a warehouse
fdw sql exec SalesWH -q "SELECT TOP 10 * FROM dbo.orders ORDER BY order_date DESC"
# Capture an estimated execution plan as SVG -- no SSMS or Windows needed
fdw sql plan SalesWH -f query.sql --format svg -o plan.svg
# -- Performance mission-control --
# See what is running right now
fdw queries running SalesWH
# Long-running queries from the past hour
fdw queries long-running SalesWH --ago 1h
# Kill a runaway session by ID
fdw queries kill SalesWH 55
# Most-repeated queries over the past 24 hours
fdw queries frequent SalesWH --ago 24h
# -- Optimize --
# Inspect a statistics histogram with inline terminal bar charts
fdw statistics show SalesWH dbo.orders st_order_date --histogram
# Re-cluster a table on a new key (transactional CTAS-swap, auto-rollback on failure)
fdw tables cluster-by SalesWH dbo.orders --cluster-by customer_id
# -- Time travel + export --
# Browse the table as it looked 2 hours ago
fdw tables read SalesWH dbo.orders --ago 2h
# Export a point-in-time snapshot to Parquet
fdw tables export SalesWH dbo.orders --output snapshot.parquet --ago 2h
# -- Governance --
# Grant SELECT on a specific table
fdw permissions sql grant SalesWH SELECT --to analyst@company.com --object dbo.orders
# Deny access to sensitive columns (column-level security)
fdw permissions cls deny SalesWH SELECT --to contractor@company.com \
--object dbo.orders --columns salary,bonus
# Create a row-level security policy (filter rows by SalesRep)
fdw permissions rls create SalesWH rls.SalesFilter \
--filter "rls.fn_sales_filter(SalesRep)" --on dbo.orders
# -- Load + scaffold --
# Load a local Parquet file and auto-create the table from its schema
fdw tables load SalesWH dbo.orders --file orders.parquet --create
# Scaffold a full dbt-fabric project wired to the warehouse
fdw dbt init SalesWH ./my-dbt-project --project-name sales_dw --with-sourcesMCP Server
Add to your MCP client configuration (e.g. Claude Desktop, VS Code):
{
"mcpServers": {
"fabric-dw": {
"command": "uvx",
"args": ["--from", "fabric-dw", "fabric-dw-mcp"]
}
}
}The MCP server exposes all CLI operations as MCP tools (workspaces, warehouses, SQL endpoints, schemas, tables, views, queries, snapshots, restore points, audit, statistics, permissions, sql-pools). Bundled Claude Code agent skills (query-optimizer, warehouse-performance, dbt-setup) are included for deeper AI-assisted analysis. Set FABRIC_AUTH in the environment if you need a non-default auth mode.
Both the skills and the MCP server install in one command via the fabric-dw plugin marketplace, for Claude Code and GitHub Copilot CLI alike: /plugin marketplace add sdebruyn/fabric-dw-mcp-cli then /plugin install fabric-dw@fabric-dw. See the Agent Skills docs for details.
Run in Docker
The Docker image's default ENTRYPOINT is the MCP server (fabric-dw-mcp). Use it as-is with your MCP client, or override the entrypoint to run the CLI instead.
docker pull ghcr.io/sdebruyn/fabric-dw:latest
# Run the MCP server (default entrypoint, connect via stdio from your MCP client):
docker run --rm -i \
-e AZURE_CLIENT_ID=β¦ \
-e AZURE_TENANT_ID=β¦ \
-e AZURE_CLIENT_SECRET=β¦ \
-e FABRIC_AUTH=sp \
ghcr.io/sdebruyn/fabric-dw
# Run the CLI instead (override the entrypoint):
docker run --rm \
--entrypoint fabric-dw \
-e AZURE_CLIENT_ID=β¦ \
-e AZURE_TENANT_ID=β¦ \
-e AZURE_CLIENT_SECRET=β¦ \
-e FABRIC_AUTH=sp \
ghcr.io/sdebruyn/fabric-dw --helpDev images (built from every main merge): ghcr.io/sdebruyn/fabric-dw:main or :<version>.dev<N>.
Package page: ghcr.io/sdebruyn/fabric-dw
Security environment variables
Variable | Default | Description |
| unset | Set to |
| unset | Set to |
| unset | Comma-separated workspace names or GUIDs the server may touch. Unset = all workspaces allowed. |
| unset | Set to |
HTTP transport
The MCP server can be started in HTTP mode for remote clients:
fabric-dw-mcp --transport http [--host 127.0.0.1] [--port 8000]It binds to loopback by default, where Host and Origin validation is handled for you. Binding anywhere else requires FABRIC_MCP_ALLOW_REMOTE=1 and --allowed-host, and the endpoint has no built-in authentication or TLS, so always front it with an authenticating reverse proxy. See Hosting the MCP server for that setup.
Develop in a container
Open the repo in GitHub Codespaces or VS Code's Remote-Containers extension. The devcontainer pre-installs Python 3.14, uv, Azure CLI, and the GitHub CLI.
Contributing
See CONTRIBUTING.md for dev setup, branch flow, and how to run tests locally.
π Docs: fdw.debruyn.dev (or run uv run --only-group docs zensical serve locally).
Security
Please report vulnerabilities privately. See SECURITY.md.
Code of Conduct
This project follows the Contributor Covenant 2.1.
License
MIT. Copyright (c) 2026 Sam Debruyn
Available Tools
123 toolsadd_audit_groupA
Add a single audit action group without overwriting the others.
Idempotent -- if the group is already present the current settings are returned unchanged. Auditing must already be enabled.
CAUTION: changes take effect immediately on the live audit policy.
CAUTION: Each audit write reads current settings via an eventually-consistent GET that may lag a recent PATCH by several minutes. Two audit writes issued within that window can cause the second to silently revert the first. Space audit writes at least a few minutes apart.
Args:
workspace: Workspace name or GUID.
warehouse: Warehouse or SQL analytics endpoint name or GUID.
group: Action group name, e.g. BATCH_COMPLETED_GROUP.
Inputs: group (required), warehouse (required), workspace (required). Returns: object
add_security_predicateA
Add a FILTER predicate to an existing row-level security policy.
Executes ALTER SECURITY POLICY ... ADD FILTER PREDICATE. There is
no predicate-type or operation parameter (#966): Fabric Data
Warehouse supports FILTER predicates only.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
policy_name: Qualified policy name ("schema.name" or "name").
fn_name: Name of the predicate function.
fn_args: Column names to pass to the predicate function.
table_schema: Schema name of the target table.
table_name: Name of the target table.
fn_schema: Schema name of the predicate function (optional -- omit
when the function lives in the default schema).
Inputs: item (required), fn_args (required), fn_name (required), workspace (required), +4 more. Returns: object
assign_workspace_to_capacityB
Assign a workspace to a Fabric capacity.
Args: workspace: Workspace name or GUID. capacity_id: UUID of the capacity to assign the workspace to.
Inputs: workspace (required), capacity_id (required). Returns: object
clear_cacheA
Erase cached workspace and item name-to-UUID mappings.
Args: scope: Which portion of the cache to clear.
- ``"workspaces"`` β clear only workspace nameβUUID entries.
- ``"items"`` β clear only item (warehouse/endpoint) entries.
- ``"all"`` (default) β clear all entries.Returns:
A dict with keys scope (the value used), workspaces_cleared
(number of workspace entries removed), and items_cleared (number
of item workspace buckets removed).
Inputs: scope. Returns: object
clear_tableA
Truncate a SQL table (remove all rows, keep structure).
Only supported on Fabric Data Warehouses (not SQL Analytics Endpoints).
The service rejects SQL Analytics Endpoints with a ToolError.
CAUTION: This is a destructive, irreversible operation. All rows will be permanently deleted. The table structure and schema are preserved. Confirm with the user before calling.
Args:
workspace: Workspace name or GUID.
item: Warehouse name or GUID. SQL Analytics Endpoints are rejected.
qualified_name: Dot-separated qualified table name, e.g. dbo.sales.
Inputs: item (required), workspace (required), qualified_name (required). Returns: object
clone_tableA
Create a zero-copy clone of a table using CREATE TABLE β¦ AS CLONE OF β¦.
Only supported on Fabric Data Warehouses (not SQL Analytics Endpoints).
Args:
workspace: Workspace name or GUID.
item: Warehouse name or GUID.
source: Qualified source table name, e.g. dbo.sales.
new_table: Qualified name for the new cloned table, e.g. dbo.sales_clone.
at: Optional ISO-8601 UTC timestamp for a point-in-time clone,
e.g. 2024-05-20T14:00:00. Must be within the data-retention
window (30 days by default). When omitted, the clone reflects the
current state of the source table.
Inputs: item (required), source (required), new_table (required), workspace (required), +1 more. Returns: object
count_table_rowsA
Return the total row count of a table via SELECT COUNT_BIG(*).
Works on both Fabric Data Warehouses and SQL Analytics Endpoints.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
qualified_name: Dot-separated qualified table name, e.g. dbo.sales.
as_of: Optional ISO-8601 UTC timestamp for a point-in-time (time-travel)
count. When supplied the query uses OPTION (FOR TIMESTAMP AS OF ...).
Omit to count the latest data.
Inputs: item (required), workspace (required), qualified_name (required), as_of. Returns: object
count_view_rowsA
Return the total row count of a view via SELECT COUNT_BIG(*).
Works on both Fabric Data Warehouses and SQL Analytics Endpoints.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
qualified_name: Dot-separated qualified view name, e.g. dbo.vw_sales.
as_of: Optional ISO-8601 UTC timestamp for a point-in-time (time-travel)
count. When supplied the query uses OPTION (FOR TIMESTAMP AS OF ...).
Omit to count the latest data.
Inputs: item (required), workspace (required), qualified_name (required), as_of. Returns: object
create_empty_tableA
Create an empty table from an explicit column spec (DDL only, no data).
Builds CREATE TABLE [schema].[table] (col TYPE [NULL|NOT NULL], β¦) from
the supplied column definitions. No data is read or inserted; this is a
pure DDL operation.
Server-side file access is unreliable in MCP deployments, so CSV/Parquet
inference is not available via this tool β use the fabric-dw tables create --from-parquet or --from-csv CLI commands instead.
Only supported on Fabric Data Warehouses (not SQL Analytics Endpoints).
When cluster_by is supplied, each column must appear in columns.
Args:
workspace: Workspace name or GUID.
item: Warehouse name or GUID.
qualified_name: Dot-separated qualified table name, e.g. dbo.sales.
columns: List of column definitions, each a dict with:
name (str) β column identifier;
sql_type (str) β Fabric-DW T-SQL type, e.g. "INT", "VARCHAR(255)";
nullable (bool, optional, default true) β whether the column allows NULL.
cluster_by: Optional list of column names for the CLUSTER BY clause
(up to 4). Each name must appear in columns.
Inputs: item (required), columns (required), workspace (required), qualified_name (required), +1 more. Returns: object
create_functionA
Create a new T-SQL user-defined function.
Scalar UDFs and inline TVFs are preview features on Fabric DW as of mid-2026. Function DDL is supported on both Data Warehouses and SQL Analytics Endpoints.
CAUTION: body is executed verbatim as DDL. Ensure the body matches the
user's intent before calling this tool.
The body should include the parameter list, RETURNS clause, and function body
(everything that follows CREATE FUNCTION [schema].[name]).
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL Analytics Endpoint name or GUID.
qualified_name: Dot-separated qualified function name, e.g. dbo.fn_clean_input.
body: The function body (parameter list, RETURNS clause, and implementation).
Inputs: body (required), item (required), workspace (required), qualified_name (required). Returns: object
create_procedureA
Create a new stored procedure.
Stored procedures are supported on both Fabric Data Warehouses and SQL Analytics Endpoints.
CAUTION: body is executed verbatim as DDL. Ensure the body
matches the user's intent before calling this tool.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
qualified_name: Dot-separated qualified procedure name, e.g. dbo.usp_load.
body: The procedure body (the AS β¦ section).
Inputs: body (required), item (required), workspace (required), qualified_name (required). Returns: object
create_restore_pointB
Create a restore point for a warehouse at the current timestamp.
Args: workspace: Workspace name or GUID. warehouse: Warehouse name or GUID. name: Optional display name (max 128 chars). description: Optional description (max 512 chars).
Inputs: warehouse (required), workspace (required), name, description. Returns: object
create_schemaA
Create a new SQL schema on a warehouse or SQL Analytics Endpoint.
Both Fabric Data Warehouses and SQL Analytics Endpoints support
CREATE SCHEMA per the Microsoft Fabric T-SQL reference.
Args: workspace: Workspace name or GUID. item: Warehouse or SQL Analytics Endpoint name or GUID. name: The schema name. Must be a valid SQL identifier.
Inputs: item (required), name (required), workspace (required). Returns: object
create_security_policyA
Create a row-level security policy.
Executes CREATE SECURITY POLICY with one or more FILTER
predicates. There is no predicate-type option (#966): Fabric Data
Warehouse supports FILTER predicates only. Each entry in predicates
must include:
fn_schema: schema of the predicate functionfn_name: name of the predicate functionfn_args: list of column names to pass to the functiontable_schema: schema of the target tabletable_name: name of the target table
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
policy_name: Qualified policy name ("schema.name" or "name").
predicates: List of predicate definitions (see above).
state: Initial policy state -- True to enable, False to disable
(default: True).
Inputs: item (required), workspace (required), predicates (required), policy_name (required), +1 more. Returns: object
create_snapshotA
Create a new warehouse snapshot.
Args: workspace: Workspace name or GUID. warehouse: Warehouse name or GUID. name: Display name for the new snapshot. description: Optional description. snapshot_dt: Optional ISO-8601 datetime string for the snapshot point-in-time. Naive datetimes (no timezone offset) are interpreted as UTC.
Inputs: name (required), warehouse (required), workspace (required), description, +1 more. Returns: object
create_sql_poolA
Add a new custom SQL pool to a workspace.
Args:
workspace: Workspace name or GUID.
name: Pool name (must be unique within the workspace).
max_percent: Max resource percentage (1-100).
is_default: Whether this pool is the default pool. Defaults to false.
optimize_for_reads: Enable read optimisation. Defaults to true.
classifier_type: Classifier type (e.g. "Application Name").
classifier_values: List of classifier values (e.g. application names).
Requires workspace admin role. This tool targets a beta / preview API.
Inputs: name (required), workspace (required), max_percent (required), is_default, +3 more. Returns: object
create_statisticsA
Create a single-column statistic on a table.
Only supported on Data Warehouses (SQL Analytics Endpoints are read-only). Only single-column statistics are supported (Fabric limitation).
Args:
workspace: Workspace name or GUID.
item: Warehouse name or GUID. SQL Analytics Endpoints are rejected.
qualified_table: Qualified table name, e.g. dbo.sales.
column: Column name to build the statistic on.
stat_name: Name for the new statistic.
fullscan: When True (default), use WITH FULLSCAN.
Ignored when sample_percent is provided.
sample_percent: Sample percentage (1-100). When provided, overrides fullscan
and uses WITH SAMPLE n PERCENT.
Inputs: item (required), column (required), stat_name (required), workspace (required), +3 more. Returns: object
create_tableA
Create a new SQL table via CTAS (CREATE TABLE AS SELECT).
Only supported on Fabric Data Warehouses (not SQL Analytics Endpoints).
The service rejects SQL Analytics Endpoints with a ToolError.
CAUTION: select_body is executed verbatim as DDL on the warehouse.
Ensure the body matches the user's intent before calling this tool.
select_body must be a single read-only SELECT or WITH (CTE)
statement. The guard is always on and fail-closed: a write keyword
(DELETE, DROP, INSERT, etc.) or a semicolon anywhere in the body is
rejected, even inside a string literal or quoted identifier. If a
legitimate query body contains a write keyword (e.g. a column alias
'DELETE'), rewrite the expression to avoid the keyword.
When cluster_by is supplied, the DDL becomes
CREATE TABLE β¦ WITH (CLUSTER BY ([c1], [c2])) AS SELECT β¦.
Column existence is not validated for CTAS because the result columns
come from the SELECT and are not known ahead of time.
Args:
workspace: Workspace name or GUID.
item: Warehouse name or GUID. SQL Analytics Endpoints are rejected.
qualified_name: Dot-separated qualified table name, e.g. dbo.sales.
select_body: Single read-only SELECT or WITH (CTE) statement for the
CTAS source. Write keywords and semicolons are rejected
fail-closed, even inside string literals or quoted identifiers.
cluster_by: Optional list of column names for the CLUSTER BY clause
(up to 4).
Inputs: item (required), workspace (required), select_body (required), qualified_name (required), +1 more. Returns: object
create_viewA
Create a new SQL view.
CAUTION: select_body is executed verbatim as DDL. Ensure the body
matches the user's intent before calling this tool.
select_body must be a single read-only SELECT or WITH (CTE)
statement. The guard is always on and fail-closed: a write keyword
(DELETE, DROP, INSERT, etc.) or a semicolon anywhere in the body is
rejected, even inside a string literal or quoted identifier. If a
legitimate view body contains a write keyword (e.g. a column alias
'DELETE'), rewrite the expression to avoid the keyword.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
qualified_name: Dot-separated qualified view name, e.g. dbo.vw_sales.
select_body: Single read-only SELECT or WITH (CTE) statement for the
view body. Write keywords and semicolons are rejected
fail-closed, even inside string literals or quoted identifiers.
Inputs: item (required), workspace (required), select_body (required), qualified_name (required). Returns: object
create_warehouseA
Create a new Warehouse in a workspace.
Args: workspace: Workspace name or GUID. name: Display name for the new warehouse. collation: Optional default collation for the new warehouse. Fabric Data Warehouse supports a fixed set of collations. Supported values include:
- ``Latin1_General_100_BIN2_UTF8`` (recommended default)
- ``Latin1_General_100_CI_AS_KS_WS_SC_UTF8``
- ``Latin1_General_CI_AS``
- ``SQL_Latin1_General_CP1_CI_AS``
When omitted, the workspace default collation is used.
Supplying an unsupported value will cause the Fabric API to
return an error. See the Fabric documentation for the full
list of supported collations.
description: Optional description for the new warehouse.
Inputs: name (required), workspace (required), collation, description. Returns: object
delete_restore_pointA
Delete a user-defined restore point.
System-created restore points cannot be deleted.
Args: workspace: Workspace name or GUID. warehouse: Warehouse name or GUID. restore_point_id: The restore point ID string.
Inputs: warehouse (required), workspace (required), restore_point_id (required). Returns: object
delete_schemaA
Drop a SQL schema from a warehouse.
CAUTION: This is a destructive, irreversible operation. The schema will
be permanently deleted. If the schema still contains tables or views,
the operation will fail unless cascade is True.
CAUTION: When cascade is True, all tables and views in the schema
are permanently deleted along with their data. Confirm explicitly with
the user before calling with cascade=True.
Both Fabric Data Warehouses and SQL Analytics Endpoints support
DROP SCHEMA per the Microsoft Fabric T-SQL reference.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL Analytics Endpoint name or GUID.
name: The schema name to drop.
cascade: When True, drop all tables and views in the schema first.
Defaults to False.
Inputs: item (required), name (required), workspace (required), cascade. Returns: object
delete_snapshotC
Delete a warehouse snapshot.
Inputs: snapshot (required), workspace (required). Returns: object
delete_sql_poolA
Delete an SQL pool from a workspace.
Args: workspace: Workspace name or GUID. pool_name: Name of the pool to delete.
Requires workspace admin role. This tool targets a beta / preview API.
Inputs: pool_name (required), workspace (required). Returns: object
delete_statisticsA
Drop a statistic via DROP STATISTICS.
CAUTION: This is a destructive, irreversible operation. Only supported on Data Warehouses (SQL Analytics Endpoints are read-only).
Args:
workspace: Workspace name or GUID.
item: Warehouse name or GUID. SQL Analytics Endpoints are rejected.
qualified_table: Qualified table name, e.g. dbo.sales.
stat_name: Name of the statistic to drop.
Inputs: item (required), stat_name (required), workspace (required), qualified_table (required). Returns: object
delete_tableA
Drop a SQL table.
Only supported on Fabric Data Warehouses (not SQL Analytics Endpoints).
The service rejects SQL Analytics Endpoints with a ToolError.
CAUTION: This is a destructive, irreversible operation. The table and all its data will be permanently deleted. Confirm with the user before calling.
Args:
workspace: Workspace name or GUID.
item: Warehouse name or GUID. SQL Analytics Endpoints are rejected.
qualified_name: Dot-separated qualified table name, e.g. dbo.sales.
Inputs: item (required), workspace (required), qualified_name (required). Returns: object
delete_warehouseC
Delete a Warehouse.
Inputs: warehouse (required), workspace (required). Returns: object
deny_permissionA
Deny permissions on a securable to a principal.
Executes DENY <permissions> ON <scope> TO <principal>.
Blocked by FABRIC_MCP_READONLY. Does NOT require
FABRIC_MCP_ALLOW_DESTRUCTIVE.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
permissions: Comma-separated permission tokens (e.g. "SELECT").
principal: Principal name to deny (Entra UPN, app GUID, or role name).
scope: Securable class -- "DATABASE" (default), "SCHEMA", or
"OBJECT".
schema: Schema name (required when scope is "SCHEMA").
object_name: Qualified object name <schema>.<object> (required when
scope is "OBJECT").
columns: Optional list of column names for column-level security
(OBJECT scope only; permissions must be SELECT, UPDATE, or
REFERENCES). Pass None (omit) for no column restriction.
Passing an empty list raises a ToolError.
Inputs: item (required), principal (required), workspace (required), permissions (required), +4 more. Returns: object
disable_auditA
Disable SQL auditing on a warehouse or SQL analytics endpoint.
CAUTION: Each audit write reads current settings via an eventually-consistent GET that may lag a recent PATCH by several minutes. Two audit writes issued within that window can cause the second to silently revert the first. Space audit writes at least a few minutes apart.
Args: workspace: Workspace name or GUID. warehouse: Warehouse or SQL analytics endpoint name or GUID.
Inputs: warehouse (required), workspace (required). Returns: object
disable_sql_poolsA
Disable custom SQL Pools for a workspace, preserving pool configuration.
Re-enabling with enable_sql_pools restores the previously saved configuration.
Requires workspace admin role. This tool targets a beta / preview API.
Inputs: workspace (required). Returns: object
drop_column_maskA
Remove a dynamic data mask from a column.
Executes ALTER TABLE ... ALTER COLUMN ... DROP MASKED.
This is a permanently destructive operation -- the mask is removed from the
column and unmasked values become visible to all users who query the column.
Requires FABRIC_MCP_ALLOW_DESTRUCTIVE=1.
Args: workspace: Workspace name or GUID. item: Warehouse or SQL endpoint name or GUID. table_schema: Schema name of the target table. table_name: Name of the target table. column_name: Name of the column whose mask to remove.
Inputs: item (required), workspace (required), table_name (required), column_name (required), +1 more. Returns: object
drop_functionA
Drop a T-SQL user-defined function.
Function DDL is supported on both Data Warehouses and SQL Analytics Endpoints.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL Analytics Endpoint name or GUID.
qualified_name: Dot-separated qualified function name, e.g. dbo.fn_clean_input.
if_exists: When true, a missing function is treated as a no-op and
{"dropped": false} is returned instead of raising an error.
Defaults to false.
Inputs: item (required), workspace (required), qualified_name (required), if_exists. Returns: object
drop_procedureA
Drop a stored procedure.
Stored procedures are supported on both Fabric Data Warehouses and SQL Analytics Endpoints.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
qualified_name: Dot-separated qualified procedure name, e.g. dbo.usp_load.
Inputs: item (required), workspace (required), qualified_name (required). Returns: object
drop_security_policyA
Drop a row-level security policy.
Executes DROP SECURITY POLICY. This is a permanently destructive
operation -- the policy and all its predicates are removed.
Requires FABRIC_MCP_ALLOW_DESTRUCTIVE=1.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
policy_name: Qualified policy name ("schema.name" or "name").
Inputs: item (required), workspace (required), policy_name (required). Returns: object
drop_security_predicateB
Drop the FILTER predicate from an existing row-level security policy.
Executes ALTER SECURITY POLICY ... DROP FILTER PREDICATE ON. The
T-SQL DROP PREDICATE ON syntax takes no operation qualifier.
There is no predicate-type parameter (#966): Fabric Data Warehouse
supports FILTER predicates only.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
policy_name: Qualified policy name ("schema.name" or "name").
table_schema: Schema name of the target table.
table_name: Name of the target table.
Inputs: item (required), workspace (required), table_name (required), policy_name (required), +1 more. Returns: object
drop_viewA
Drop a SQL view.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
qualified_name: Dot-separated qualified view name, e.g. dbo.vw_sales.
Inputs: item (required), workspace (required), qualified_name (required). Returns: object
enable_auditA
Enable SQL auditing on a warehouse or SQL analytics endpoint.
CAUTION: The pre-flight GET used to round-trip the existing action-group list is eventually consistent and may lag a recent PATCH by several minutes. If the action-group list was changed within that window, this call may silently revert it. Space audit writes at least a few minutes apart.
Args: workspace: Workspace name or GUID. warehouse: Warehouse or SQL analytics endpoint name or GUID. retention_days: Log retention in days (0-3650; 0 = unlimited). Default 0.
Inputs: warehouse (required), workspace (required), retention_days. Returns: object
enable_sql_poolsB
Enable custom SQL Pools for a workspace without modifying pool definitions.
Requires workspace admin role. This tool targets a beta / preview API.
Inputs: workspace (required). Returns: object
execute_sqlA
Execute an arbitrary SQL statement or batch against a warehouse or SQL Analytics Endpoint.
Prefer dedicated tools for common operations: use read_table or read_view to fetch rows, count_table_rows or count_view_rows to count, list_tables, list_views, list_schemas to discover objects, get_table_columns or get_view_columns to inspect schemas, and delete_table, rename_table, or clear_table to mutate. Dedicated tools return structured, typed results with no dialect pitfalls or batch-truncation surprises.
WARNING: this tool executes arbitrary SQL against the target. DDL (DROP,
ALTER, TRUNCATE) and DML (DELETE, UPDATE) are permitted unless
FABRIC_MCP_READONLY=1 is set. Use only when the user explicitly
requests data modification. Default to SELECT when the user's intent is
read-only investigation.
Supports both Warehouse and SQL Analytics Endpoint items. Multi-statement
batches are allowed; the tool keeps the last result set that has a column
list (i.e. the last query), not the last statement. A batch that ends with
DDL/DML after a query, such as SELECT id FROM t; UPDATE t SET x = 1;,
returns the SELECT result and does not separately report that the
UPDATE ran. A statement with no result set at all (DDL/DML with nothing
else in the batch) returns columns=[] and rows=[].
datetime and Decimal column values are pre-serialised to strings.
bytes / varbinary columns are base64-encoded and their column names are
suffixed with __base64.
For large tables, add a TOP clause or WHERE predicate to the query rather
than relying solely on max_rows. The driver fetches at most
max_rows + 1 rows (enough to detect truncation) so memory is bounded,
but pushing the limit into the query itself is always more efficient.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL Analytics Endpoint name or GUID.
query: SQL statement or batch to execute.
max_rows: Maximum rows to return (1-10000, default 1000). When the
result set is larger the response includes "truncated": true.
Returns:
A dict with keys columns (list[str]), rows (list[list[Any]]),
rowcount (int; -1 when the driver does not report a count),
row_count_returned (int), and truncated (bool).
Inputs: item (required), query (required), workspace (required), max_rows. Returns: object
generate_dbt_profileA
Generate dbt-fabric project file contents for a Fabric Data Warehouse.
Returns the generated file contents as text strings. Because the MCP server cannot write to the caller's local filesystem, it is the caller's responsibility to write the returned strings to the appropriate files.
Authentication note: dbt-fabric is Entra-only. ServicePrincipal mode emits
{{ env_var(...) }} placeholders for tenant_id / client_id / client_secret
β no literal secrets are included in the output.
Args:
workspace: Workspace name or GUID.
item: Warehouse name or GUID.
project_name: dbt project name (default: sanitized warehouse name).
profile_name: dbt profile name (default: same as project_name).
schema: Default schema (default: dbo).
target: dbt output target name (default: dev).
threads: Number of dbt threads (default: 4).
authentication: dbt-fabric authentication string β
auto (DefaultAzureCredential), CLI (interactive),
or ServicePrincipal. Defaults to the server's auth mode.
with_sources: When True, generate a _sources.yml from the
warehouse's actual schemas and tables.
Returns:
A dict with keys:
- profiles_yml: content for profiles.yml.
- dbt_project_yml: content for dbt_project.yml.
- sources_yml: content for models/staging/_sources.yml.
- requirements_txt: content for requirements.txt.
- gitignore: content for .gitignore.
Inputs: item (required), workspace (required), schema, target, +5 more. Returns: object
get_audit_settingsA
Fetch the current SQL audit settings for a warehouse or SQL analytics endpoint.
Args: workspace: Workspace name or GUID. warehouse: Warehouse or SQL analytics endpoint name or GUID.
Inputs: warehouse (required), workspace (required). Returns: object
get_cluster_columnsA
Return the data-clustering columns of a table, ordered by clustering ordinal.
Only supported on Fabric Data Warehouses. SQL Analytics Endpoints raise a
ToolError. Returns an empty list when no clustering columns are defined.
Args:
workspace: Workspace name or GUID.
item: Warehouse name or GUID.
qualified_name: Dot-separated qualified table name, e.g. dbo.sales.
Inputs: item (required), workspace (required), qualified_name (required). Returns: result
get_functionA
Fetch the full definition of a T-SQL user-defined function (schema.fn).
Returns the function definition (from sys.sql_modules) and its parameter list
(from sys.parameters). Scalar UDFs and inline TVFs are supported on both
Data Warehouses and SQL Analytics Endpoints.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL Analytics Endpoint name or GUID.
qualified_name: Dot-separated qualified function name, e.g. dbo.fn_clean_input.
Inputs: item (required), workspace (required), qualified_name (required). Returns: object
get_procedureA
Fetch the full definition of a stored procedure (schema.proc).
Stored procedures are supported on both Fabric Data Warehouses and SQL Analytics Endpoints.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
qualified_name: Dot-separated qualified procedure name, e.g. dbo.usp_load.
Inputs: item (required), workspace (required), qualified_name (required). Returns: object
get_query_planA
Capture the estimated SHOWPLAN_XML execution plan for a SQL query without executing it.
This tool does NOT execute the query β it only retrieves the estimated execution
plan as SHOWPLAN_XML. Because no data is modified, this tool is permitted even
under FABRIC_MCP_READONLY=1.
The plan XML uses the standard namespace
http://schemas.microsoft.com/sqlserver/2004/07/showplan and can be opened
in SSMS, Azure Data Studio, or uploaded to pastetheplan.com for visual analysis.
Since the query is not executed, DDL/DML query text is safe to plan without modifying any data.
Supports both Warehouse and SQL Analytics Endpoint items.
Format options:
"xml"(default, backwards-compatible) β returns the raw SHOWPLAN_XML string inplan_xml. Existing callers relying on{"plan_xml": str}continue to work unchanged."tree"β parses the XML into a native nested list of dicts (one entry per statement) inplan. Best for agent reasoning over the plan structure."json"β same tree, serialised to an indented JSON string inplan_json. Ready to write out or pass through as compact text."mermaid"β renders a Mermaidflowchart TDdiagram string inmermaid. Paste into mermaid.live or embed in GitHub Markdown.
Artifact formats (SVG/HTML/DOT) are CLI-only. They write files to disk and
are only available via fdw sql plan --format <fmt> -o <file>. The MCP
server never writes files (ambiguous cwd, invisible side-effects).
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL Analytics Endpoint name or GUID.
query: SQL statement to generate an estimated execution plan for.
format: Output format β one of "xml" (default), "tree",
"json", or "mermaid".
Returns: A dict whose shape depends on format:
- ``xml`` β ``{"format": "xml", "plan_xml": str}``
- ``tree`` β ``{"format": "tree", "plan": list[dict]}``
- ``json`` β ``{"format": "json", "plan_json": str}``
- ``mermaid`` β ``{"format": "mermaid", "mermaid": str}``
Inputs: item (required), query (required), workspace (required), format. Returns: object
get_request_detailA
Look up a completed query from queryinsights.exec_requests_history.
Uses distributed_statement_id to retrieve full query text and execution metrics after the query completes.
Args: workspace: Workspace name or GUID. item: Warehouse or SQL Analytics Endpoint name or GUID. dist_statement_id: The GUID identifying the distributed statement to look up.
Inputs: item (required), workspace (required), dist_statement_id (required). Returns: result
get_restore_pointA
Return a single restore point by ID.
Args:
workspace: Workspace name or GUID.
warehouse: Warehouse name or GUID.
restore_point_id: The restore point ID string (e.g. "1726617378000").
Inputs: warehouse (required), workspace (required), restore_point_id (required). Returns: object
get_sql_endpointA
Return details for a single SQL analytics endpoint (name or GUID).
Inputs: endpoint (required), workspace (required). Returns: object
get_sql_poolA
Return details for a single SQL pool by name.
Args: workspace: Workspace name or GUID. pool_name: The pool name.
Requires workspace admin role. This tool targets a beta / preview API.
Inputs: pool_name (required), workspace (required). Returns: object
get_sql_pools_statusA
Return whether custom SQL Pools are enabled for a workspace.
Returns only the workspace-level enabled/disabled switch
(customSQLPoolsEnabled). Use list_sql_pools to see the pool
list, or get_sql_pool for a single pool's details.
Requires workspace admin role. This tool targets a beta / preview API endpoint that may change before general availability.
Inputs: workspace (required). Returns: object
get_table_columnsB
Return column metadata for a SQL table via sys.columns.
Works on both Fabric Data Warehouses and SQL Analytics Endpoints.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
qualified_name: Dot-separated qualified table name, e.g. dbo.sales.
Inputs: item (required), workspace (required), qualified_name (required). Returns: result
get_table_health_metricsA
Return health metrics for a table via sp_get_table_health_metrics.
Only supported on SQL Analytics Endpoints (not Data Warehouses). The proc surfaces Delta/Parquet layout issues such as small files, fragmentation, excessive deletes/updates, and delayed checkpoints.
The stored procedure is Generally Available (announced at Build 2026) but its output column schema is not yet documented by Microsoft. Columns and rows are passed through verbatim.
Args:
workspace: Workspace name or GUID.
item: SQL Analytics Endpoint name or GUID. Data Warehouses are
rejected with a ToolError.
qualified_name: Dot-separated qualified table name, e.g. dbo.sales.
Inputs: item (required), workspace (required), qualified_name (required). Returns: object
get_viewB
Fetch the full definition of a view (schema.view).
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
qualified_name: Dot-separated qualified view name, e.g. dbo.vw_sales.
Inputs: item (required), workspace (required), qualified_name (required). Returns: object
get_view_columnsA
Return column metadata for a SQL view via sys.columns.
Works on both Fabric Data Warehouses and SQL Analytics Endpoints.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
qualified_name: Dot-separated qualified view name, e.g. dbo.vw_sales.
Inputs: item (required), workspace (required), qualified_name (required). Returns: result
get_warehouseC
Return details for a single warehouse (name or GUID).
Inputs: warehouse (required), workspace (required). Returns: object
get_warehouse_settingsA
Return the current server-side database settings for a warehouse.
Reads result_set_caching, time_travel_retention_days,
time_travel_retention_cutoff_date, and data_lake_log_publishing
from sys.databases.
Both Data Warehouses and SQL Analytics Endpoints are supported.
Args: workspace: Workspace name or GUID. item: Warehouse or SQL Analytics Endpoint name or GUID.
Inputs: item (required), workspace (required). Returns: object
get_workspaceB
Return details for a single workspace (name or GUID).
Inputs: workspace (required). Returns: object
grant_permissionA
Grant permissions on a securable to a principal.
Executes GRANT <permissions> ON <scope> TO <principal>.
Blocked by FABRIC_MCP_READONLY. Does NOT require
FABRIC_MCP_ALLOW_DESTRUCTIVE.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
permissions: Comma-separated permission tokens (e.g. "SELECT,INSERT").
principal: Grantee principal name (Entra UPN, app GUID, or role name).
scope: Securable class -- "DATABASE" (default), "SCHEMA", or
"OBJECT".
schema: Schema name (required when scope is "SCHEMA").
object_name: Qualified object name <schema>.<object> (required when
scope is "OBJECT").
with_grant_option: When True, allows the grantee to grant the
permission to others (adds WITH GRANT OPTION).
columns: Optional list of column names for column-level security
(OBJECT scope only; permissions must be SELECT, UPDATE, or
REFERENCES). Pass None (omit) for no column restriction.
Passing an empty list raises a ToolError.
Inputs: item (required), principal (required), workspace (required), permissions (required), +5 more. Returns: object
import_table_from_urlA
Load data into an existing Data Warehouse table via COPY INTO from a remote URL.
The target table must already exist and have a compatible schema.
For auto-create with schema inference from local files, use the CLI
tables load --file --create command instead.
if_exists controls behaviour when the table already exists:
"fail"(default): raise an error if the table already exists, or if it does not exist (the table must be created first with create_empty_table or create_table)."append": load rows into the existing table without modification. Raises an error if the table does not exist."truncate": TRUNCATE the existing table, then load, both inside a single transaction so a failed load leaves the existing rows intact (atomic replace). RequiresFABRIC_MCP_ALLOW_DESTRUCTIVE=1. Raises an error if the table does not exist."replace": not supported for remote URLs (schema inference requires downloading the file). Use"truncate"to keep the current schema, or download locally and use the CLI with--create --if-exists replace.
Supported file types: CSV, PARQUET. JSON remote URLs require
downloading and converting locally first; use the CLI tables load
command for local files (including JSON).
For OneLake or same-tenant URLs, no credential is needed. For secured
external URLs supply credential_type and the appropriate
secret/identity values.
CAUTION: truncate is permanently destructive.
Confirm the source URL and target table before calling.
Note: secret / identity values are accepted but are NEVER logged
or included in any debug output.
Args:
workspace: Workspace name or GUID.
item: Warehouse name or GUID. SQL Analytics Endpoints are rejected.
qualified_name: Dot-separated qualified table name, e.g. dbo.sales.
url: Source URL (OneLake DFS URL or external Azure Blob URL).
file_type: CSV or PARQUET.
if_exists: Policy when the target table already exists.
credential_type: Credential type for the source URL.
secret: Credential secret (not logged).
identity: Identity for managed-identity or service-principal.
delimiter: CSV column delimiter.
has_header: Whether the CSV file has a header row.
encoding: CSV file encoding.
field_quote: CSV field-quote character.
row_terminator: CSV row terminator.
max_errors: Maximum errors before aborting.
rejected_row_location: URL for rejected-row output.
Inputs: url (required), item (required), file_type (required), workspace (required), +12 more. Returns: object
kill_sessionA
Terminate a session on a warehouse by session_id.
Args: workspace: Workspace name or GUID. item: Warehouse or SQL Analytics Endpoint name or GUID. session_id: Session ID to terminate (must be a positive integer).
Inputs: item (required), workspace (required), session_id (required). Returns: object
list_capabilitiesA
List all available MCP tools grouped by domain.
Call this tool first to discover what dedicated tools are available
before falling back to execute_sql. Dedicated tools return typed,
structured results and avoid SQL dialect pitfalls.
Returns: A dict mapping domain name to a sorted list of tool names in that domain. The dict itself is sorted by domain key.
Inputs: no named parameters. Returns: object
list_capacitiesA
List all Fabric capacities the caller has access to.
Requires the Capacity.Read.All permission. Returns a 403
ToolError when the caller lacks that permission.
Inputs: no named parameters. Returns: result
list_connectionsB
Return all active SQL connections on a warehouse or SQL Analytics Endpoint.
Args: workspace: Workspace name or GUID. item: Warehouse or SQL Analytics Endpoint name or GUID.
Inputs: item (required), workspace (required). Returns: result
list_database_principalsA
List database principals from sys.database_principals.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
principal_type: Filter by type -- "user" for users, "role" for
database roles, "all" or omit for no filter.
Inputs: item (required), workspace (required), principal_type. Returns: result
list_frequent_queriesA
Return frequently-run queries from queryinsights.frequently_run_queries.
Args: workspace: Workspace name or GUID. item: Warehouse or SQL Analytics Endpoint name or GUID. limit: Maximum rows to return (1-10000, default 100). since: Optional ISO-8601 lower bound on last_run_start_time. until: Optional ISO-8601 upper bound on last_run_start_time.
Inputs: item (required), workspace (required), limit, since, +1 more. Returns: result
list_functionsA
List T-SQL user-defined functions on a warehouse or SQL Analytics Endpoint.
Scalar UDFs (FN) and inline TVFs (IF) are preview features on Fabric DW as of mid-2026. Function DDL is supported on both Data Warehouses and SQL Analytics Endpoints.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL Analytics Endpoint name or GUID.
schema: When provided, only functions in this schema are returned.
kind: Filter by function kind β "scalar" (FN only),
"inline-tvf" (IF only), or "all" (FN + IF + TF, the default).
Inputs: item (required), workspace (required), kind, schema. Returns: result
list_item_permissionsA
Return principals with access to a Warehouse or SQL Analytics Endpoint item.
Uses the Fabric admin API. Accepts both Data Warehouses and SQL Analytics Endpoints -- the item kind is resolved automatically from its GUID.
Requires Fabric Administrator role (admin API).
See https://learn.microsoft.com/en-us/fabric/admin/microsoft-fabric-admin for details.
Args: workspace: Workspace name or GUID. item: Warehouse or SQL endpoint name or GUID.
Inputs: item (required), workspace (required). Returns: result
list_locksA
Return active lock rows from sys.dm_tran_locks joined with sys.dm_exec_requests.
Args: workspace: Workspace name or GUID. item: Warehouse or SQL Analytics Endpoint name or GUID. limit: Maximum rows to return (1-10000, default 100). waiting_only: When True, restrict to locks with request_status WAIT or CONVERT. blocked_only: When True, show only blocked sessions (victims). The blocker's session_id appears in blocking_session_id. include_database: When True, include DATABASE-scoped lock rows (excluded by default).
Inputs: item (required), workspace (required), limit, blocked_only, +2 more. Returns: result
list_long_running_queriesB
Return long-running queries from queryinsights.long_running_queries.
Args: workspace: Workspace name or GUID. item: Warehouse or SQL Analytics Endpoint name or GUID. limit: Maximum rows to return (1-10000, default 100). since: Optional ISO-8601 lower bound on last_run_start_time. until: Optional ISO-8601 upper bound on last_run_start_time.
Inputs: item (required), workspace (required), limit, since, +1 more. Returns: result
list_masked_columnsA
List columns with dynamic data masking from sys.masked_columns.
Returns all masked columns on the target Data Warehouse or SQL Analytics Endpoint. Filter by table_schema and/or table_name to narrow results.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
table_schema: Optional schema filter (case-insensitive). Pass None
to include all schemas.
table_name: Optional table name filter (case-insensitive). Pass None
to include all tables.
Inputs: item (required), workspace (required), table_name, table_schema. Returns: result
list_proceduresA
List stored procedures on a warehouse or SQL Analytics Endpoint.
Stored procedures are supported on both Fabric Data Warehouses and SQL Analytics Endpoints.
Args: workspace: Workspace name or GUID. item: Warehouse or SQL endpoint name or GUID. schema: When provided, only procedures in this schema are returned.
Inputs: item (required), workspace (required), schema. Returns: result
list_request_historyA
Return completed SQL requests from queryinsights.exec_requests_history.
Args: workspace: Workspace name or GUID. item: Warehouse or SQL Analytics Endpoint name or GUID. limit: Maximum rows to return (1-10000, default 100). since: Optional ISO-8601 lower bound on submit_time. until: Optional ISO-8601 upper bound on submit_time.
Inputs: item (required), workspace (required), limit, since, +1 more. Returns: result
list_restore_pointsB
Return all restore points for a warehouse.
Inputs: warehouse (required), workspace (required). Returns: result
list_running_queriesA
Return all currently-executing queries on a warehouse or SQL Analytics Endpoint.
Args: workspace: Workspace name or GUID. item: Warehouse or SQL Analytics Endpoint name or GUID.
Inputs: item (required), workspace (required). Returns: result
list_schemasA
List user-defined SQL schemas on a warehouse or SQL Analytics Endpoint.
System schemas (sys, INFORMATION_SCHEMA, db_* fixed-role
schemas, guest) are excluded. dbo is included as it is
user-writable.
Listing schemas is a read-only operation and works on both Fabric Data Warehouses and SQL Analytics Endpoints.
Args: workspace: Workspace name or GUID. item: Warehouse or SQL Analytics Endpoint name or GUID.
Inputs: item (required), workspace (required). Returns: result
list_security_policiesA
List row-level security policies from sys.security_policies.
Returns all security policies and their predicates for the target Data Warehouse or SQL Analytics Endpoint.
Args: workspace: Workspace name or GUID. item: Warehouse or SQL endpoint name or GUID.
Inputs: item (required), workspace (required). Returns: result
list_session_historyA
Return completed sessions from queryinsights.exec_sessions_history.
Args: workspace: Workspace name or GUID. item: Warehouse or SQL Analytics Endpoint name or GUID. limit: Maximum rows to return (1-10000, default 100). since: Optional ISO-8601 lower bound on session_start_time. until: Optional ISO-8601 upper bound on session_start_time.
Inputs: item (required), workspace (required), limit, since, +1 more. Returns: result
list_snapshotsB
Return all snapshots belonging to a warehouse.
Inputs: warehouse (required), workspace (required). Returns: result
list_sql_endpointsA
List all SQL analytics endpoints in a workspace.
Args:
workspace: Workspace name or GUID. Optional when all_workspaces
is True; required otherwise.
all_workspaces: When True, ignore workspace and aggregate
results across every workspace the caller can see.
Inputs: workspace, all_workspaces. Returns: result
list_sql_permissionsA
List T-SQL database permissions from sys.database_permissions.
Reads from sys.database_permissions joined to sys.database_principals. Returns DATABASE, SCHEMA, and OBJECT class securables with readable names.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
principal: Filter by principal name (optional).
schema: Filter by schema name -- returns SCHEMA class rows for this
schema (optional).
object_name: Filter by qualified object name <schema>.<object>
-- returns OBJECT class rows for this object (optional).
Inputs: item (required), workspace (required), schema, principal, +1 more. Returns: result
list_sql_pool_insightsA
Return SQL pool insight events from queryinsights.sql_pool_insights.
Args: workspace: Workspace name or GUID. warehouse: Warehouse or SQL Analytics Endpoint name or GUID. limit: Maximum rows to return (1-10000, default 100). since: Optional ISO-8601 lower bound on timestamp. until: Optional ISO-8601 upper bound on timestamp.
Inputs: warehouse (required), workspace (required), limit, since, +1 more. Returns: result
list_sql_poolsA
Return the list of custom SQL pools for a workspace.
Requires workspace admin role. This tool targets a beta / preview API.
Inputs: workspace (required). Returns: result
list_statisticsA
List statistics on a warehouse or SQL Analytics Endpoint.
Both Data Warehouses and SQL Analytics Endpoints are supported.
Args: workspace: Workspace name or GUID. item: Warehouse or SQL endpoint name or GUID. schema: When provided, only statistics on tables in this schema are returned. table: When provided, only statistics on this table (unqualified name) are returned. user_only: When True, only user-created statistics are returned. auto_only: When True, only auto-created statistics are returned.
Inputs: item (required), workspace (required), table, schema, +2 more. Returns: result
list_tablesA
List SQL tables on a warehouse or SQL Analytics Endpoint.
Args: workspace: Workspace name or GUID. item: Warehouse or SQL endpoint name or GUID. schema: When provided, only tables in this schema are returned.
Inputs: item (required), workspace (required), schema. Returns: result
list_table_sync_statusA
Show per-table metadata sync freshness via sys.dm_db_external_tables_log_status.
Only supported on SQL Analytics Endpoints (not Data Warehouses), and only
on endpoints created after the workspace's 'New metadata sync' (preview)
setting was enabled. A table with no matching DMV row still appears, with
its sync fields null instead of the row being dropped -- this means no
sync information is available for that table, NOT that it has never
synced; do not treat a null row as proof the table has never synced.
IMPORTANT for agents: by default this listing only includes tables
already present in the endpoint's catalog (sys.tables), which is
itself maintained by the metadata sync. A Lakehouse table whose
discovery has not completed, or has failed, has no catalog row and so
is absent from this result entirely -- it will not appear as a row
with empty fields, it simply will not be there. Do not conclude a
table does not exist, or was deleted, just because it is missing from
this list.
Pass check_lakehouse=True to close part of that gap: it
cross-references the backing Lakehouse's own table inventory (one or
more extra REST calls -- avoid setting this on every call of a tool an
agent may invoke repeatedly) and adds a row with
in_endpoint_catalog=false and all sync fields null for every
table it finds there but not in the catalog, comparing names exactly
(case-sensitively, matching Fabric's default collation). A Lakehouse
table that differs from a catalog table only by case gets
case_mismatched_catalog_name set to that catalog name instead of
being reported as a flat miss.
This works for both classic and schema-enabled Lakehouse-backed
endpoints (the schema-enabled path uses a preview OneLake table API).
If the endpoint's backing item cannot be resolved to a Lakehouse at
all (a mirrored database, or similar), check_lakehouse=True
raises a ToolError explaining why, rather than silently returning
the unchanged catalog-only result: a caller that explicitly asked for
this cross-check must never read "no extra rows" as "fully
discovered". check_lakehouse=True also cannot be combined with
schema or table (raises a ToolError): it always compares the
whole endpoint. If check_lakehouse=True fails or an expected table
is still missing after trying it, call
refresh_sql_endpoint_metadata (or tell the user to run
fdw sql-endpoints refresh) to force an item-level sync, then check
again.
Args:
workspace: Workspace name or GUID.
item: SQL Analytics Endpoint name or GUID. Data Warehouses are
rejected with a ToolError.
schema: When provided, only tables in this schema are returned.
table: When provided, filter to this single (bare, unqualified)
table name. Requires schema to also be given.
check_lakehouse: When True, cross-reference the backing
Lakehouse's table inventory for tables missing from the
endpoint catalog entirely. See above for its limits. Mutually
exclusive with schema and table.
Inputs: item (required), workspace (required), table, schema, +1 more. Returns: result
list_viewsA
List SQL views on a warehouse or SQL Analytics Endpoint.
Args: workspace: Workspace name or GUID. item: Warehouse or SQL endpoint name or GUID. schema: When provided, only views in this schema are returned.
Inputs: item (required), workspace (required), schema. Returns: result
list_warehousesA
List all warehouses and SQL analytics endpoints in a workspace.
Args:
workspace: Workspace name or GUID. Optional when all_workspaces
is True; required otherwise.
all_workspaces: When True, ignore workspace and aggregate
results across every workspace the caller can see.
Inputs: workspace, all_workspaces. Returns: result
list_workspacesA
List all Fabric workspaces the caller has access to.
When a workspace allowlist is configured (via FABRIC_MCP_WORKSPACES
env var or [mcp] workspace_allowlist in config.toml) only the
workspaces that match the allowlist (by name or GUID) are returned.
Inputs: no named parameters. Returns: result
load_table_from_urlA
Load data into a Data Warehouse table via COPY INTO from a remote URL.
Supported file types: CSV, PARQUET. JSON remote URLs require
downloading and converting locally first; use the CLI tables load
command for local files (including JSON).
For OneLake or same-tenant URLs, no credential is needed. For secured
external URLs (Azure Blob Storage SAS, etc.), supply credential_type
and the appropriate secret/identity values.
CAUTION: This operation loads data into the target table. Confirm the source URL and target table before calling.
Note: secret / identity values are accepted but are NEVER logged
or included in any debug output.
Args:
workspace: Workspace name or GUID.
item: Warehouse name or GUID. SQL Analytics Endpoints are rejected.
qualified_name: Dot-separated qualified table name, e.g. dbo.sales.
url: Source URL (OneLake DFS URL or external Azure Blob URL).
file_type: CSV or PARQUET.
credential_type: Credential type for the source URL.
secret: Credential secret (not logged).
identity: Identity for managed-identity or service-principal.
delimiter: CSV column delimiter.
has_header: Whether the CSV file has a header row.
encoding: CSV file encoding.
field_quote: CSV field-quote character.
row_terminator: CSV row terminator.
max_errors: Maximum errors before aborting.
rejected_row_location: URL for rejected-row output.
Inputs: url (required), item (required), file_type (required), workspace (required), +11 more. Returns: object
my_permissionsA
Return permissions for the current connection via sys.fn_my_permissions.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
scope: Scope string -- "database" (default), "schema:<name>",
or "object:<schema>.<object>".
Inputs: item (required), workspace (required), scope. Returns: result
read_tableA
Return up to count rows from a table as JSON-serialisable columns + rows.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
qualified_name: Dot-separated qualified table name, e.g. dbo.sales.
count: Maximum number of rows to return (1-10000, default 10).
as_of: Optional ISO-8601 UTC timestamp for a point-in-time (time-travel)
read. When supplied the query uses OPTION (FOR TIMESTAMP AS OF ...).
Omit to read the latest data.
Inputs: item (required), workspace (required), qualified_name (required), as_of, +1 more. Returns: object
read_viewA
Return up to count rows from a view as JSON-serialisable columns + rows.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
qualified_name: Dot-separated qualified view name, e.g. dbo.vw_sales.
count: Maximum number of rows to return (1-10000, default 10).
as_of: Optional ISO-8601 UTC timestamp for a point-in-time (time-travel)
read. When supplied the query uses OPTION (FOR TIMESTAMP AS OF ...).
Omit to read the latest data.
Inputs: item (required), workspace (required), qualified_name (required), as_of, +1 more. Returns: object
refresh_sql_endpoint_metadataA
Refresh metadata for a SQL analytics endpoint (sync from the underlying Lakehouse).
This is a long-running operation (LRO) that is polled to completion.
Returns a list of per-table sync results. Use this tool for SCHEMA
changes (tables added or dropped). For cheap, per-table DATA-only
staleness on a table you already know exists, use
refresh_table_metadata instead -- it does not pick up schema
changes.
Args:
workspace: Workspace name or GUID.
endpoint: SQL analytics endpoint name or GUID.
recreate_tables: When True, drop and recreate all tables during
the refresh. Use to resolve inconsistencies or force a clean
rebuild. Destructive β use with caution. Requires
FABRIC_MCP_ALLOW_DESTRUCTIVE=1 when enabled.
Inputs: endpoint (required), workspace (required), recreate_tables. Returns: result
refresh_table_metadataA
Refresh one table's metadata via sys.sp_dw_refresh_ext_table.
This is the cheap, per-table refresh for DATA-only staleness: it
re-reads the table's underlying Delta log without a full item-level
sync. Use refresh_sql_endpoint_metadata instead when the SCHEMA
changed (tables added or dropped) -- this tool does not pick up
schema changes.
Only supported on SQL Analytics Endpoints (not Data Warehouses), and
only on endpoints created after the workspace's 'New metadata sync'
(preview) setting was enabled. Mutating (respects
FABRIC_MCP_READONLY) but NOT destructive -- it never drops or
recreates anything, so it does not require the
FABRIC_MCP_ALLOW_DESTRUCTIVE opt-in.
qualified_name must already exist in the endpoint's catalog --
the procedure does not create it, and raises a ToolError naming
the table instead. That same error also fires when the table exists
but you lack permission to it, since the driver does not
distinguish the two -- refresh_sql_endpoint_metadata only helps
the missing case, not a permissions one. The procedure may also
decline a table by type, for reasons Microsoft does not document,
raised as a ToolError in the same family as the legacy-sync one
above; that failure also points at refresh_sql_endpoint_metadata
(or fdw sql-endpoints refresh) as the working alternative.
Args:
workspace: Workspace name or GUID.
item: SQL Analytics Endpoint name or GUID. Data Warehouses are
rejected with a ToolError.
qualified_name: Dot-separated qualified table name, e.g.
dbo.sales.
Inputs: item (required), workspace (required), qualified_name (required). Returns: object
remove_audit_groupA
Remove a single audit action group without overwriting the others.
Idempotent -- if the group is not present the current settings are returned unchanged. Auditing must already be enabled.
CAUTION: changes take effect immediately on the live audit policy.
CAUTION: Each audit write reads current settings via an eventually-consistent GET that may lag a recent PATCH by several minutes. Two audit writes issued within that window can cause the second to silently revert the first. Space audit writes at least a few minutes apart.
Args:
workspace: Workspace name or GUID.
warehouse: Warehouse or SQL analytics endpoint name or GUID.
group: Action group name, e.g. BATCH_COMPLETED_GROUP.
Inputs: group (required), warehouse (required), workspace (required). Returns: object
rename_snapshotD
Rename a warehouse snapshot.
Inputs: new_name (required), snapshot (required), workspace (required), description. Returns: object
rename_tableA
Rename a SQL table via sp_rename (Data-Warehouse-only).
Renames the table in-place within the same schema using T-SQL
EXEC sp_rename. Both the current qualified name and the new bare
name are passed as bound parameters β no SQL injection is possible.
sp_rename cannot move a table to a different schema, so new_name
must be an unqualified (bare) name without a dot.
Args:
workspace: Workspace name or GUID.
item: Warehouse name or GUID. SQL Analytics Endpoints are rejected.
qualified_name: Current dot-separated qualified table name, e.g.
dbo.sales.
new_name: New table name (unqualified, e.g. sales_v2). Must not
contain a dot.
Inputs: item (required), new_name (required), workspace (required), qualified_name (required). Returns: object
rename_viewA
Rename a SQL view via sp_rename.
Works on both Data Warehouses and SQL Analytics Endpoints.
The new name must be a bare (unqualified) identifier β sp_rename
cannot move a view across schemas.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
qualified_name: Current dot-separated qualified view name,
e.g. dbo.vw_sales.
new_name: New bare view name (no schema prefix), e.g. vw_revenue.
Inputs: item (required), new_name (required), workspace (required), qualified_name (required). Returns: object
rename_warehouseB
Rename a Warehouse (and optionally update its description).
Inputs: new_name (required), warehouse (required), workspace (required), description. Returns: object
restore_warehouse_in_placeA
Restore a warehouse in-place to a restore point.
WARNING: This is a destructive, long-running operation. The warehouse will be unavailable for approximately 10 minutes while the restore completes.
Args: workspace: Workspace name or GUID. warehouse: Warehouse name or GUID. restore_point_id: The restore point ID string to restore to.
Inputs: warehouse (required), workspace (required), restore_point_id (required). Returns: object
revoke_permissionA
Revoke permissions on a securable from a principal.
Executes REVOKE <permissions> ON <scope> FROM <principal>.
Blocked by FABRIC_MCP_READONLY. Requires
FABRIC_MCP_ALLOW_DESTRUCTIVE=1 because revoke removes an existing
permission (destructive operation).
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
permissions: Comma-separated permission tokens (e.g. "SELECT,INSERT").
principal: Principal name to revoke from (Entra UPN, app GUID, or role name).
scope: Securable class -- "DATABASE" (default), "SCHEMA", or
"OBJECT".
schema: Schema name (required when scope is "SCHEMA").
object_name: Qualified object name <schema>.<object> (required when
scope is "OBJECT").
columns: Optional list of column names for column-level security
(OBJECT scope only; permissions must be SELECT, UPDATE, or
REFERENCES). Pass None (omit) for no column restriction.
Passing an empty list raises a ToolError.
grant_option_only: When True, revokes only the grant option (adds
GRANT OPTION FOR), leaving the base permission in place.
cascade: When True, cascades the revocation to principals the
grantee has granted the permission to (adds CASCADE).
Inputs: item (required), principal (required), workspace (required), permissions (required), +6 more. Returns: object
roll_snapshot_timestampA
Roll a snapshot's timestamp forward (or reset to current).
Args: workspace: Workspace name or GUID. warehouse: Parent warehouse name or GUID (used for the SQL connection). snapshot_name: The snapshot database name to roll. new_dt: Optional ISO-8601 datetime string; defaults to CURRENT_TIMESTAMP. Naive datetimes (no timezone offset) are interpreted as UTC.
Inputs: warehouse (required), workspace (required), snapshot_name (required), new_dt. Returns: object
set_audit_action_groupsA
Replace the audited action groups for a warehouse or SQL analytics endpoint.
Only replaces the action groups. Does not toggle the audit enabled or disabled state; if auditing was Disabled before the call it remains Disabled afterwards.
CAUTION: Each audit write reads current settings via an eventually-consistent GET that may lag a recent PATCH by several minutes. The retention period read from that GET is round-tripped; if retention was changed within the lag window, this call may silently revert it. Space audit writes at least a few minutes apart.
Args: workspace: Workspace name or GUID. warehouse: Warehouse or SQL analytics endpoint name or GUID. action_groups: List of audit action group names.
Inputs: warehouse (required), workspace (required), action_groups (required). Returns: object
set_audit_retentionA
Update the audit log retention period without changing the audit enabled/disabled state.
Audit must already be enabled; if disabled, enable it first with enable_audit.
CAUTION: The pre-flight GET used to round-trip the existing action-group list is eventually consistent and may lag a recent PATCH by several minutes. If the action-group list was changed within that window, this call may silently revert it. Space audit writes at least a few minutes apart.
Args: workspace: Workspace name or GUID. warehouse: Warehouse or SQL analytics endpoint name or GUID. days: Retention period in days (1-3650). The API enforces its own upper bound.
Inputs: days (required), warehouse (required), workspace (required). Returns: object
set_cluster_columnsA
Change (or remove) the data-clustering columns of an existing table.
Rebuilds the table via a transactional CTAS-swap:
CREATE TABLE [schema].[__recluster_<hex>] [WITH (CLUSTER BY (...))] AS SELECT * FROM [schema].[orig]DROP TABLE [schema].[orig]EXEC sp_renameto restore the original name
All three steps run inside ONE transaction. Any failure rolls back automatically β no orphan temp table is left behind.
CAUTION: This operation copies the full table (runtime is proportional
to table size). Dependent views and stored procedures that reference
this table by name are NOT automatically updated by sp_rename and
may need refreshing after the swap.
Only supported on Fabric Data Warehouses (not SQL Analytics Endpoints).
Args:
workspace: Workspace name or GUID.
item: Warehouse name or GUID. SQL Analytics Endpoints are rejected.
qualified_name: Dot-separated qualified table name, e.g. dbo.sales.
cluster_by: New list of column names for the CLUSTER BY clause
(up to 4). Pass null or an empty list to remove clustering
(rebuilds table without CLUSTER BY).
Inputs: item (required), workspace (required), qualified_name (required), cluster_by. Returns: object
set_column_maskA
Apply or replace a dynamic data mask on a column.
Executes ALTER TABLE ... ALTER COLUMN ... ADD MASKED WITH (FUNCTION = '...').
ADD MASKED replaces any existing mask on the column without error.
Blocked by FABRIC_MCP_READONLY.
Supported mask function types:
"default"-- full masking; no extra args."email"-- email masking (exposes first char and".com"suffix); no extra args."random"-- numeric random mask; requires start and end."partial"-- custom string partial mask; requires prefix, padding, and suffix.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
table_schema: Schema name of the target table.
table_name: Name of the target table.
column_name: Name of the column to mask.
fn_type: Mask function type -- "default", "email", "random",
or "partial" (case-insensitive).
start: Lower bound for random() masking (required when fn_type is
"random"). Must be <= end.
end: Upper bound for random() masking (required when fn_type is
"random").
prefix: Leading characters to expose for partial() masking (required
when fn_type is "partial").
padding: Replacement padding string for partial() masking (required
when fn_type is "partial"). Must not contain ", ),
;, --, control characters (including U+0085, U+2028,
U+2029), and must not exceed 128 characters.
suffix: Trailing characters to expose for partial() masking (required
when fn_type is "partial").
Inputs: item (required), fn_type (required), workspace (required), table_name (required), +7 more. Returns: object
set_data_lake_log_publishingA
Enable or disable Delta Lake log publishing on a warehouse.
Executes ALTER DATABASE CURRENT SET DATA_LAKE_LOG_PUBLISHING = { AUTO | PAUSED }
and returns the effective settings read back after the change.
Only supported on Fabric Data Warehouses (not SQL Analytics Endpoints).
SQL Analytics Endpoints are rejected with a ToolError.
Args:
workspace: Workspace name or GUID.
item: Warehouse name or GUID. SQL Analytics Endpoints are rejected.
enabled: True to enable Delta Lake log publishing (= AUTO),
False to disable it (= PAUSED).
Inputs: item (required), enabled (required), workspace (required). Returns: object
set_result_set_cachingA
Enable or disable result-set caching on a warehouse.
Executes ALTER DATABASE CURRENT SET RESULT_SET_CACHING { ON | OFF }
and returns the effective settings read back after the change.
Only supported on Fabric Data Warehouses (not SQL Analytics Endpoints).
SQL Analytics Endpoints are rejected with a ToolError.
Args:
workspace: Workspace name or GUID.
item: Warehouse name or GUID. SQL Analytics Endpoints are rejected.
enabled: True to enable result-set caching, False to disable it.
Inputs: item (required), enabled (required), workspace (required). Returns: object
set_security_policy_stateA
Enable or disable a row-level security policy.
Executes ALTER SECURITY POLICY ... WITH (STATE = ON|OFF).
Not destructive -- enabling or disabling a policy is reversible.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
policy_name: Qualified policy name ("schema.name" or "name").
enabled: True to enable the policy, False to disable it.
Inputs: item (required), enabled (required), workspace (required), policy_name (required). Returns: object
set_time_travel_retentionA
Set the time-travel retention period on a warehouse.
Executes ALTER DATABASE CURRENT SET TIME_TRAVEL_RETENTION_PERIOD = <n> DAYS
and returns the effective settings read back after the change.
Only supported on Fabric Data Warehouses (not SQL Analytics Endpoints).
SQL Analytics Endpoints are rejected with a ToolError.
Args: workspace: Workspace name or GUID. item: Warehouse name or GUID. SQL Analytics Endpoints are rejected. days: Retention period in days. Must be in the range 1-120 (inclusive).
Inputs: days (required), item (required), workspace (required). Returns: object
set_workspace_collationA
Set the default Data Warehouse collation for a workspace.
Args: workspace: Workspace name or GUID. collation: Collation to apply. Fabric Data Warehouse supports a fixed set of collations. Supported values include:
- ``Latin1_General_100_BIN2_UTF8`` (recommended default)
- ``Latin1_General_100_CI_AS_KS_WS_SC_UTF8``
- ``Latin1_General_CI_AS``
- ``SQL_Latin1_General_CP1_CI_AS``
Supplying an unsupported value will cause the Fabric API to
return an error. See the Fabric documentation for the full
list of supported collations.
Inputs: collation (required), workspace (required). Returns: object
show_statisticsA
Show details of a statistic using DBCC SHOW_STATISTICS.
Returns the stat header, density vector, and histogram steps. Both Data Warehouses and SQL Analytics Endpoints are supported.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
qualified_table: Qualified table name, e.g. dbo.sales.
stat_name: The name of the statistic to show.
histogram_only: When True, return only the histogram steps.
Inputs: item (required), stat_name (required), workspace (required), qualified_table (required), +1 more. Returns: object
takeover_warehouseC
Take ownership of a Warehouse.
Inputs: warehouse (required), workspace (required). Returns: object
transfer_functionA
Move a T-SQL user-defined function to another schema via ALTER SCHEMA TRANSFER.
Function DDL is supported on both Data Warehouses and SQL Analytics Endpoints -- unlike table transfer, no endpoint guard applies here.
CAUTION: ALTER SCHEMA ... TRANSFER does not rewrite the schema name
inside the function's stored definition (sys.sql_modules.definition).
After a transfer, get_function may still show the old schema name
in the CREATE ... AS header, even though the function now lives in
the new schema. This tool does not rewrite the definition text.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL Analytics Endpoint name or GUID.
qualified_name: Current dot-separated qualified function name, e.g.
dbo.fn_clean_input.
target_schema: Schema to move the function into, e.g. archive.
Inputs: item (required), workspace (required), target_schema (required), qualified_name (required). Returns: object
transfer_procedureA
Move a stored procedure to another schema via ALTER SCHEMA ... TRANSFER OBJECT::....
Stored procedures are supported on both Fabric Data Warehouses and
SQL Analytics Endpoints; unlike transfer_table, no endpoint guard
is applied here.
CAUTION: ALTER SCHEMA ... TRANSFER moves the procedure but does
NOT rewrite the schema name inside its stored definition. After a
transfer, get_procedure may still show the OLD schema name in the
CREATE ... AS header even though the procedure now lives in the
new schema.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
qualified_name: Current dot-separated qualified procedure name,
e.g. dbo.usp_load.
target_schema: Schema to move the procedure into, e.g. archive.
Inputs: item (required), workspace (required), target_schema (required), qualified_name (required). Returns: object
transfer_tableA
Move a SQL table to another schema via ALTER SCHEMA ... TRANSFER OBJECT::....
Data-Warehouse-only: transferring a table between schemas via T-SQL is
not supported on the Fabric SQL Analytics Endpoint and can break the
OneLake sync, so SQL Analytics Endpoints are rejected with a ToolError.
CAUTION: Permissions granted directly on the table are dropped by the engine when the schema changes. Dependent views and stored procedures that reference the table by its old schema-qualified name are NOT automatically updated and may need refreshing after the transfer.
Args:
workspace: Workspace name or GUID.
item: Warehouse name or GUID. SQL Analytics Endpoints are rejected.
qualified_name: Current dot-separated qualified table name, e.g.
dbo.sales.
target_schema: Schema to move the table into, e.g. archive.
Inputs: item (required), workspace (required), target_schema (required), qualified_name (required). Returns: object
transfer_viewA
Move a SQL view to another schema via ALTER SCHEMA ... TRANSFER OBJECT::....
Works on both Data Warehouses and SQL Analytics Endpoints β no DW-only guard is applied.
CAUTION: ALTER SCHEMA ... TRANSFER moves the view but does not
rewrite the schema name inside the view's stored definition
(sys.sql_modules.definition, OBJECT_DEFINITION()). After a
transfer, the returned (and any subsequent get_view) definition
may still show the old schema name in the CREATE ... AS header,
even though the view now lives in the new schema. This tool does not
rewrite the definition text β doing so would require parsing and
regenerating SQL, which this project deliberately avoids.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
qualified_name: Current dot-separated qualified view name,
e.g. dbo.vw_sales.
target_schema: Schema to move the view into, e.g. archive.
Inputs: item (required), workspace (required), target_schema (required), qualified_name (required). Returns: object
update_functionA
Redefine a T-SQL user-defined function via CREATE OR ALTER FUNCTION.
Note: ALTER FUNCTION cannot change the function kind (e.g. scalar to inline TVF). The body must be compatible with the original function's kind.
Scalar UDFs and inline TVFs are preview features on Fabric DW as of mid-2026. Function DDL is supported on both Data Warehouses and SQL Analytics Endpoints.
CAUTION: body is executed verbatim as DDL. Ensure the body matches the
user's intent before calling this tool.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL Analytics Endpoint name or GUID.
qualified_name: Dot-separated qualified function name, e.g. dbo.fn_clean_input.
body: The new function body (parameter list, RETURNS clause, and implementation).
Inputs: body (required), item (required), workspace (required), qualified_name (required). Returns: object
update_procedureA
Redefine a stored procedure via CREATE OR ALTER PROCEDURE.
Stored procedures are supported on both Fabric Data Warehouses and SQL Analytics Endpoints.
CAUTION: body is executed verbatim as DDL. Ensure the body
matches the user's intent before calling this tool.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
qualified_name: Dot-separated qualified procedure name, e.g. dbo.usp_load.
body: The new procedure body (the AS β¦ section).
Inputs: body (required), item (required), workspace (required), qualified_name (required). Returns: object
update_restore_pointA
Rename and/or update the description of a restore point.
At least one of name or description must be provided.
Args: workspace: Workspace name or GUID. warehouse: Warehouse name or GUID. restore_point_id: The restore point ID string. name: New display name (max 128 chars). description: New description (max 512 chars).
Inputs: warehouse (required), workspace (required), restore_point_id (required), name, +1 more. Returns: object
update_sql_poolA
Update an existing SQL pool. Only the parameters you supply are changed.
Args: workspace: Workspace name or GUID. name: Name of the pool to update. max_percent: New max resource percentage (1-100), or omit to keep current. is_default: Set or clear the default flag, or omit to keep current. optimize_for_reads: Enable/disable read optimisation, or omit to keep current. classifier_type: New classifier type, or omit to keep current. classifier_values: New classifier value list, or omit to keep current.
Requires workspace admin role. This tool targets a beta / preview API.
Inputs: name (required), workspace (required), is_default, max_percent, +3 more. Returns: object
update_statisticsA
Update an existing statistic via UPDATE STATISTICS.
Only supported on Data Warehouses (SQL Analytics Endpoints are read-only).
Args:
workspace: Workspace name or GUID.
item: Warehouse name or GUID. SQL Analytics Endpoints are rejected.
qualified_table: Qualified table name, e.g. dbo.sales.
stat_name: Name of the statistic to update.
fullscan: When True (default), use WITH FULLSCAN.
Ignored when sample_percent is provided.
sample_percent: Sample percentage (1-100). When provided, overrides fullscan
and uses WITH SAMPLE n PERCENT.
Inputs: item (required), stat_name (required), workspace (required), qualified_table (required), +2 more. Returns: object
update_viewA
Redefine a SQL view via CREATE OR ALTER VIEW.
CAUTION: select_body is executed verbatim as DDL. Ensure the body
matches the user's intent before calling this tool.
select_body must be a single read-only SELECT or WITH (CTE)
statement. The guard is always on and fail-closed: a write keyword
(DELETE, DROP, INSERT, etc.) or a semicolon anywhere in the body is
rejected, even inside a string literal or quoted identifier. If a
legitimate view body contains a write keyword (e.g. a column alias
'DELETE'), rewrite the expression to avoid the keyword.
Args:
workspace: Workspace name or GUID.
item: Warehouse or SQL endpoint name or GUID.
qualified_name: Dot-separated qualified view name, e.g. dbo.vw_sales.
select_body: Single read-only SELECT or WITH (CTE) statement for the
new view body. Write keywords and semicolons are rejected
fail-closed, even inside string literals or quoted identifiers.
Inputs: item (required), workspace (required), select_body (required), qualified_name (required). Returns: object
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelogβ¦
TDQS
Scored across 123 tools
Most tools target a distinct resource and action, with detailed descriptions that help an agent choose correctly. However, load_table_from_url and import_table_from_url are near-duplicates (both COPY INTO from a URL), and execute_sql overlaps with virtually every dedicated tool. Some related clusters (list_warehouses vs. list_sql_endpoints) also require careful reading to avoid misselection.
The dominant pattern is verb_noun and is largely readable, but destructive operations are split inconsistently between 'drop_' (drop_view, drop_procedure, drop_function) and 'delete_' (delete_table, delete_warehouse, delete_schema). The same conceptual operation is also named both load_table_from_url and import_table_from_url, breaking a strict one-verb-per-action convention.
At 123 tools, this is far beyond the range where an agent can efficiently discover and select the right tool without significant overhead. Even a comprehensive warehouse management surface could be consolidated into far fewer, broader tools. The extreme count will likely cause prompt-size and selection-cost problems.
The surface is exhaustive for the stated domain: full lifecycle coverage for warehouses, SQL endpoints, tables, views, procedures, functions, schemas, snapshots, restore points, SQL pools, permissions, security policies, masking, auditing, and query monitoring. Any residual gap (e.g. row-level DML or ALTER TABLE) is covered by the deliberately explicit execute_sql fallback. There are no obvious dead ends.
Maintenance
Related MCP Connectors
Official Microsoft MCP Server to query Microsoft Entra data using natural language
MCP server for mandates, delegation, policy-gated execution, credential grants, and audit.
Governed data discovery, exact queries, decisions, simulations, and runtime utilities over MCP.
Hosted SQL Server MCP server for Claude, ChatGPT and Cursor, with each action switched per account.
Related MCP Servers
- FlicenseNot gradedqualityDmaintenanceMCP server for reading and analyzing Fabric semantic models. Supports getting model definitions and executing DAX queries against Power BI datasets.17-
- AlicenseBqualityCmaintenanceMCP server for Microsoft Fabric REST APIs that enables data engineers and analysts to manage Fabric components using AI assistants.219AGPL 3.0
- FlicenseCqualityDmaintenanceA Python MCP server that lets you manage Microsoft Fabric through natural language in Claude Code or Claude Desktop, with 77+ tools covering workspaces, lakehouses, warehouses, SQL, DAX, semantic models, notebooks, pipelines, OneLake, and Microsoft Graph.83-
- AlicenseNot gradedqualityCmaintenanceMCP server for managing Microsoft Fabric notebooks and Dataflow Gen2 via the Fabric REST API. Supports reading, creating, updating, running, and deleting items within workspaces.MIT