DataForge Semantic MCP Server
DataForge Semantic MCP Server gives AI agents complete MCP access to DataForge's semantic analytics layer — reading business definitions and evolving them through 65 tools (24 read-only, 41 write).
Read semantic context: list projects and versions; get measures, dimensions, facts, RMD, and consolidated RMD exports.
Generate SQL: produce SQL for data marts and measures, and inspect physical data mart views (nothing is executed or stored).
Explore the data model: list/get fact tables, dimension groups, relationships, connections, and cached connection schemas.
Manage RMD content: create, replace, update, and delete measures, dimensions, and facts, including bulk writes.
Manage dimension groups and fact tables: create/delete groups, set or remove group dimensions, create/delete fact tables, and assign/unassign elements.
Manage filters and relationships: write and delete verification filters and star-schema relationships.
Manage projects and versions: create, update, and delete projects/versions; set the global version; transfer ownership.
Control access: grant, change, or revoke project access for users.
Version transfer: export versions to Git or file, dry-run check/preview imports, and import from Git or file.
Manage Git connections: create, update, delete, and test saved Git connections.
Operational tasks: check server health and refresh cached RMD snapshots.
Provides a read-only semantic gateway to the DataForge Product API by Business Qlik, enabling AI agents to browse projects and versions, and retrieve normalized metadata including measures, dimensions, and full Relationship Metadata (RMD).
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., "@DataForge Semantic MCP Server list all measures and dimensions for project 392 version 948"
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.
DataForge Semantic MCP Server
Give your AI agent the business meaning behind your analytics — and the power to change it.
Quick start · Tools · Writing data · Open WebUI · Documentation
THIS SERVER IS NOT READ-ONLY — IT CAN DELETE YOUR DATA
38 of its 65 tools create, overwrite and permanently delete DataForge content: projects, versions, measures, dimensions, facts, dimension groups, fact tables, relationships, access grants and Git connections.
Tool | What it does |
| Deletes a project with every version and all of its content |
| Overwrites an entire version |
| Resets every optional field you do not pass |
There is no kill switch in this server.
The only thing between an LLM agent and your production semantic model is the
effective project role of the API key in DATAFORGE_API_KEY.
➜ Want read-only? Use a read-only key.
Issue a key whose project role is analyst or viewer. Every write endpoint requires
developer or above and answers 403 DF_API.WRITE_ACCESS_DENIED — the API enforces
this, this server does not. Point agents at a non-production version first.
Every write lands in the DataForge audit log with before/after snapshots.
Related MCP server: FabricIQ MCP Server
Why
DataForge holds the business definitions behind your analytics: what "Total Revenue" actually means, which database column feeds it, how customers are grouped. Without that context an AI agent is guessing.
This server hands the agent that context — structured, normalized and cached — and gives it a safe, idempotent way to evolve the model.
What You Get
65 MCP tools — 24 read-only, 41 that change state (38 of them write to DataForge).
Read — 24 tools
Group | Tools |
Discovery |
|
Semantic layer |
|
SQL generation |
|
Data model |
|
Data marts |
|
Connections |
|
Full export |
|
Access & Git |
|
Ops |
|
Write — 41 tools ⚠️
Group | Tools |
RMD content |
|
Dimension groups |
|
Fact tables |
|
Filters & links |
|
Projects & versions |
|
Access |
|
Version transfer |
|
Git connections |
|
Local state |
|
Full reference: read tools · data model · write tools
Quick Start
1 · Install
pip install -e ".[dev]"2 · Configure
cp .env.example .envDATAFORGE_BASE_URL=https://api.prod-df.businessqlik.com
DATAFORGE_API_KEY=your_api_key_hereDATAFORGE_BASE_URL is the root of the API, not of the site — requests go to
<base>/df-api/v2/…. Installations that serve the API behind a prefix need it spelled out
(https://dataforge.example.com/api); point it at the site root and the web app answers
with HTML and HTTP 200, which the server reports as DATAFORGE_INVALID_RESPONSE.
The key decides what an agent can destroy. analyst / viewer → effectively
read-only. developer and above → can delete measures, versions and whole projects.
3 · Run
python -m dataforge_mcp # stdio, for Claude Desktop / Cursor
docker compose up # Streamable HTTP on http://localhost:8080/mcpTransport |
| Clients |
stdio |
| Claude Desktop, Cursor |
Streamable HTTP |
| Open WebUI, MCP Inspector, remote deployments |
HTTP+SSE |
| Deprecated — kept for existing deployments |
Over HTTP the port is the only thing between the network and the 38 write tools, so set
MCP_AUTH_TOKEN for any bind that is not loopback. Full reference:
docs/api/transports.md.
Or skip MCP entirely and use it as a library:
import asyncio
from dataforge_mcp import create_semantic_service
async def main():
service = create_semantic_service()
projects = await service.list_projects()
project_id = projects["projects"][0]["id"]
versions = await service.list_versions(project_id=project_id)
version_id = next(v["id"] for v in versions["versions"] if v["is_global"])
rmd = await service.get_rmd(project_id=project_id, version_id=version_id)
print(f"{rmd['stats']['measure_count']} measures, "
f"{rmd['stats']['dimension_count']} dimensions, "
f"{rmd['stats']['fact_count']} facts")
asyncio.run(main())Claude Desktop
{
"mcpServers": {
"dataforge": {
"command": "python",
"args": ["-m", "dataforge_mcp"],
"env": {
"DATAFORGE_BASE_URL": "https://api.prod-df.businessqlik.com",
"DATAFORGE_API_KEY": "your_api_key_here"
}
}
}
}Before you paste this: the key above decides whether an agent can delete your
projects. Use an analyst / viewer key for read-only work. Write tools carry
readOnlyHint=false and destructiveHint=true, so a well-behaved client asks you to
confirm — but that confirmation belongs to the client, not to this server.
Then ask Claude: "What measures are available in the Fashion Retail project?" — it will
chain df_list_projects → df_list_versions → df_get_measures on its own.
Open WebUI
Open WebUI speaks Streamable HTTP only, which this server now serves natively — no
mcpo proxy needed.
cp .env.example .env
# set DATAFORGE_API_KEY, and MCP_AUTH_TOKEN:
# python -c "import secrets; print(secrets.token_urlsafe(32))"
docker compose up -dThen in Open WebUI, as an administrator: Settings → Admin → Integrations → + Add
Connection, type MCP (Streamable HTTP), URL http://host.docker.internal:8080/mcp,
Auth Bearer with your MCP_AUTH_TOKEN. All 65 tools appear.
Every Open WebUI user who can use the tools gets all 65, including the 38 that delete
data — and they all share one DataForge key. Start with ananalyst / viewer key and a
non-production version.
Full walkthrough, URL table and troubleshooting: docs/api/open-webui.md.
Examples
Read, then write
Every measure, dimension and fact carries a stable id. That id is exactly what the write
tools take, so a row you just read goes straight back:
measures = await service.get_measures(project_id=392, version_id=948)
target = next(m for m in measures["measures"] if m["name"] == "Total revenue")
await service.write_measure(
project_id=392, version_id=948,
mode="update", # PATCH — touches only what you pass
measure_id=int(target["id"]),
measure_description="Gross revenue across all channels",
)
mode="replace"is a PUT: it resets every optional field you omit. Useupdateunless you mean a full overwrite.
SQL for a measure
measures = await service.get_measures(project_id=392, version_id=948, include_sql=True)
for m in measures["measures"]:
if m.get("sql_code"):
for script in m["sql_code"]["sql_scripts"]:
print(f"{m['name']} @ {script['fact_table_name']}:\n {script['sql']}")SQL for a data mart
A generation failure comes back as data, not an error — that is what separates "this mart cannot produce SQL right now" from "this mart does not exist":
marts = await service.list_data_marts(project_id=392, version_id=948, search="Sales")
mart_id = int(marts["data_marts"][0]["id"])
result = await service.generate_sql(
project_id=392, version_id=948, data_mart_id=mart_id, limit=100
)
if result["succeeded"]:
print(result["target_db_type"], result["sql_script"]) # postgres | clickhouse | sqlserver
else:
for err in result["validation_errors"]:
print("cannot generate:", err["message"])Explore the star schema
tables = await service.list_fact_tables(project_id=392, version_id=948)
for ft in tables["fact_tables"]:
print(f"{ft['name']}: {ft['measures_count']}M / {ft['dimensions_count']}D")
rels = await service.list_relationships(project_id=392, version_id=948)
for r in rels["relationships"]:
print(f"{r['source_fact_table']['name']} -> "
f"{r['target_dimension_group']['name']} ({r['relationship_type']})")Writing Data
Every write generates an Idempotency-Key (UUID v4); you can also supply your own. Within
24 hours, repeating a POST with the same key replays the original response instead of
applying the change twice.
That is what makes a retry after a 5xx safe. A POST sent without a key is never
retried, and a timeout on one is reported with possibly_applied: true so you verify
before resending.
Bulk and assignment calls apply each item in its own transaction. When some are rejected, the result says so explicitly:
result = await service.bulk_write_measures(
project_id=392, version_id=948,
measures=[
{"measure_name": "Net revenue", "measure_type": "Base"}, # created
{"id": "1000", "measure_description": "Updated"}, # updated
],
)
if result["status"] == "partial":
for failure in result["failed"]:
print(failure["index"], failure["error"]["code"])Operation | Effective project role |
All reads | any access to the project |
RMD and data model writes |
|
Project access management | project owner or company administrator |
Version export, import dry runs |
|
Version import |
|
Git connection management | company administrator |
A project the key cannot see answers 404, never 403 — existence is never disclosed.
The v2 error envelope is surfaced in full, including which field was wrong and why:
{
"error": {
"code": "DATAFORGE_INVALID_SOURCE_TABLE",
"api_code": "invalid_source_table",
"message": "Table not found in the connection schema",
"fields": [{ "field": "connected_source.table", "code": "invalid_value" }],
"hint": "Call df_get_connection_schema to list the tables cached for this connection.",
"retryable": false
}
}Local validation produces the same shape, so the agent never learns two formats.
Reads are cache-first with TTL and a last-known-good fallback: if the API goes down you still get the last successful response.
Writes do the opposite — they never read the cache, never serve stale data, and drop the whole scope they touched. A read right after a write always hits the API.
Troubleshooting
Symptom | Cause | Fix |
Tools fail with |
| Add the API prefix, usually |
| Installed from a tree older than 0.3.1, where | Reinstall from 0.3.1 or newer |
| MCP SDK version mismatch | This server needs |
| An argument is missing, misspelled, or a type that cannot be coerced | Read |
On Windows, the endpoint 404s although the container started cleanly | Git Bash rewrote | Prefix the command with |
Every request gets |
| Unset it, or give it a real value |
More, per area: configuration.md · transports.md
Documentation
Measures, dimensions, facts, RMD | |
Data marts, connections, fact tables, relationships | |
Everything that modifies DataForge | |
Source objects, pagination, error catalogue | |
Environment, Docker, architecture | |
stdio, Streamable HTTP, auth, Origin validation, proxies | |
Step-by-step connection guide | |
Why the transport works the way it does | |
What changed per release, and what breaks |
Key Features
Complete DF API v2 surface — reads, writes, version transfer, Git connection registry
Three transports — stdio, Streamable HTTP (Open WebUI-ready, with bearer auth and Origin validation), and legacy SSE
Safety annotations — every tool carries MCP
readOnlyHint/destructiveHint, so clients can warn before a destructive callIdempotent writes — automatic
Idempotency-Keymakes retries safeSmart caching — TTL + last-known-good for reads; scope invalidation for writes
Actionable errors —
api_code, offendingfields[]and a fix hintRate-limit aware — 100 req/60 s per key, with
retry_after_secondsCredential hygiene — API keys and Git tokens never reach logs or tool output
Library-first — usable from plain Python, no MCP server required
Development
pip install -e ".[dev]"
pytest # 502 tests
pytest tests/test_mcp_server.py -v # end-to-end over MCP, stdio path
pytest tests/test_mcp_streamable_http.py -v # end-to-end over MCP, HTTP path
ruff check src/ tests/
ruff format src/ tests/Fixtures in tests/fixtures/api_v2.py are transcribed verbatim from the DataForge API
documentation. When the API changes, that file is what gets updated first — inventing
fixtures is how a v1-shaped test suite once hid five real defects.
License
Proprietary. For use with the DataForge Product API.
Available Tools
65 toolsdf_assign_to_fact_tableA
WRITES TO DATAFORGE. Attach existing measures, dimensions, facts or dimension groups to a fact table. Ids are applied in order; already-assigned or unknown ids come back in failed[] with status=partial rather than failing the whole call.
Inputs: project_id (required), version_id (required), element_ids (required), element_type (required), +2 more.
df_bulk_write_dimensionsB
WRITES TO DATAFORGE. Create and/or update many dimensions in one call.
Inputs: dimensions (required), project_id (required), version_id (required), idempotency_key.
df_bulk_write_factsB
WRITES TO DATAFORGE. Create and/or update many facts in one call.
Inputs: facts (required), project_id (required), version_id (required), idempotency_key.
df_bulk_write_measuresA
WRITES TO DATAFORGE. Create and/or update many measures in one call. An item with id is updated, an item without one is created. Items are applied in order, each in its own transaction: a partial result comes back with status=partial and a failed[] array addressing rows by index.
Inputs: measures (required), project_id (required), version_id (required), idempotency_key.
df_check_import_sourceBIdempotent
Dry run: validate an import source without writing anything. Returns valid, errors[], warnings[] and element counts.
Inputs: project_id (required), version_id (required), path, branch, +7 more.
df_create_git_connectionA
WRITES TO DATAFORGE. Register a Git connection for the company. Requires a company administrator API key. Credentials are stored encrypted and never returned.
Inputs: name (required), branch (required), platform (required), authentication (required), +4 more.
df_create_projectB
WRITES TO DATAFORGE. Create a project. An initial version is created with it.
Inputs: name (required), color, description, idempotency_key.
df_create_versionA
WRITES TO DATAFORGE. Create a project version. Content is cloned from clone_from_version, or from the current global version when omitted. Counts against the licence version limit.
Inputs: name (required), project_id (required), is_global, idempotency_key, +1 more.
df_delete_dimensionADestructiveIdempotent
PERMANENTLY DELETES DATA IN DATAFORGE. Delete a dimension. Rejected if it belongs to a dimension group or is referenced by a formula.
Inputs: project_id (required), version_id (required), dimension_id (required).
df_delete_dimension_groupADestructiveIdempotent
PERMANENTLY DELETES DATA IN DATAFORGE. Delete a dimension group. Rejected while it is assigned to a fact table.
Inputs: project_id (required), version_id (required), dimension_group_id (required).
df_delete_factBDestructiveIdempotent
PERMANENTLY DELETES DATA IN DATAFORGE. Delete a fact.
Inputs: fact_id (required), project_id (required), version_id (required).
df_delete_fact_tableADestructiveIdempotent
PERMANENTLY DELETES DATA IN DATAFORGE. Delete a fact table. Rejected while it has active relationships.
Inputs: project_id (required), version_id (required), fact_table_id (required).
df_delete_git_connectionADestructiveIdempotent
PERMANENTLY DELETES DATA IN DATAFORGE. Delete a saved Git connection.
Inputs: connection_id (required).
df_delete_measureADestructiveIdempotent
PERMANENTLY DELETES DATA IN DATAFORGE. Delete a measure. Rejected with a conflict if another element's formula references it.
Inputs: measure_id (required), project_id (required), version_id (required).
df_delete_projectADestructiveIdempotent
PERMANENTLY DELETES DATA IN DATAFORGE. Delete a project WITH ALL OF ITS VERSIONS and their entire content. This cannot be undone.
Inputs: project_id (required).
df_delete_relationshipADestructiveIdempotent
PERMANENTLY DELETES DATA IN DATAFORGE. Delete a relationship.
Inputs: project_id (required), version_id (required), relationship_id (required).
df_delete_verification_filterBDestructiveIdempotent
PERMANENTLY DELETES DATA IN DATAFORGE. Delete a verification filter.
Inputs: filter_id (required), project_id (required), version_id (required), fact_table_id.
df_delete_versionADestructiveIdempotent
PERMANENTLY DELETES DATA IN DATAFORGE. Delete a version and all of its content. The current global version cannot be deleted.
Inputs: project_id (required), version_id (required).
df_export_version_to_fileA
WRITES TO DATAFORGE. Export a version to a .dfexport.zip archive and return a signed download link. Nothing in DataForge changes, but the archive is stored.
Inputs: project_id (required), version_id (required), options.
df_export_version_to_gitA
WRITES TO DATAFORGE. Export a version's configuration to a Git repository. Pass either connection_id (a saved Git connection) or authentication. Exporting an unchanged version creates no commit and returns commit_hash=null.
Inputs: branch (required), project_id (required), version_id (required), commit_message (required), +6 more.
df_generate_sqlARead-onlyIdempotent
Generate the SQL query of a data mart. Nothing is executed and nothing is stored. A generation failure comes back as succeeded=false with validation_errors, not as an error.
Inputs: project_id (required), version_id (required), data_mart_id (required), limit, +2 more.
df_get_connectionBRead-onlyIdempotent
Get one connection: host, port, database, schema and username. Set include_db_schema=true to get the full cached table/column schema.
Inputs: project_id (required), version_id (required), connection_id (required), language, +2 more.
df_get_connection_schemaARead-onlyIdempotent
Get the cached schema of a connection (tables and columns). This is a snapshot taken when the connection was last refreshed, not a live query. Use it to pick valid table and column names for write operations.
Inputs: project_id (required), version_id (required), connection_id (required), language, +1 more.
df_get_consolidated_rmdBRead-onlyIdempotent
Get the full raw export of a project version: RMD content plus dimension groups, fact tables and relationships in one payload.
Inputs: project_id (required), version_id (required), language, use_cache, +1 more.
df_get_data_martBRead-onlyIdempotent
Get one data mart in full: source fact tables, selected measures, facts and dimensions with their aggregation and filter settings.
Inputs: project_id (required), version_id (required), data_mart_id (required), language, +1 more.
df_get_data_mart_viewARead-onlyIdempotent
Get the physical view materialized for a data mart: existence, object type, database, status, staleness and last refresh.
Inputs: project_id (required), version_id (required), data_mart_id (required), language, +1 more.
df_get_dimension_groupARead-onlyIdempotent
Get one dimension group: primary key, member dimensions with their hierarchy levels, and the fact tables related to it.
Inputs: project_id (required), version_id (required), dimension_group_id (required), language, +1 more.
df_get_dimensionsBRead-onlyIdempotent
Get all dimensions of a project version. Each dimension carries a stable id and, where bound to a database, a connected_source naming its connection, table and column.
Inputs: project_id (required), version_id (required), language, use_cache.
df_get_factsCRead-onlyIdempotent
Get all facts of a project version.
Inputs: project_id (required), version_id (required), language, use_cache.
df_get_fact_tableARead-onlyIdempotent
Get one fact table: assigned measures, dimensions, facts, dimension groups and verification filters. Set include_dependencies=true to get the formula dependency tree of each measure.
Inputs: project_id (required), version_id (required), fact_table_id (required), language, +2 more.
df_get_git_connectionBRead-onlyIdempotent
Get one saved Git connection.
Inputs: connection_id (required), use_cache.
df_get_measuresARead-onlyIdempotent
Get all measures (business metrics) of a project version, paged through automatically. Each measure carries a stable id that the write tools accept. Set include_sql=true to also get generated SQL per measure.
Inputs: project_id (required), version_id (required), language, use_cache, +1 more.
df_get_project_accessARead-onlyIdempotent
List the owner and every user with explicit access to a project.
Inputs: project_id (required), language, use_cache.
df_get_relationshipARead-onlyIdempotent
Get one relationship with its foreign and primary key source objects.
Inputs: project_id (required), version_id (required), relationship_id (required), language, +1 more.
df_get_rmdBRead-onlyIdempotent
Get the normalized semantic context of a project version: project, version, measures, dimensions, facts and counts. Shares one API call and one cache entry with df_get_consolidated_rmd.
Inputs: project_id (required), version_id (required), language, use_cache, +2 more.
df_healthARead-onlyIdempotent
Check server health, DataForge API connectivity and cache status.
Inputs: no named parameters.
df_import_version_from_fileBDestructiveIdempotent
PERMANENTLY DELETES DATA IN DATAFORGE. Import a version from a local .dfexport.zip archive. On-premises installations only. target_method=replace OVERWRITES the version in the path.
Inputs: file_path (required), project_id (required), version_id (required), target_version_name (required), +5 more.
df_import_version_from_gitADestructiveIdempotent
PERMANENTLY DELETES DATA IN DATAFORGE. Import a version from Git. target.method=create makes a new version; target.method=replace OVERWRITES THE VERSION IN THE PATH entirely. Run df_preview_import first. conflict_strategy=overwrite also applies deletions.
Inputs: branch (required), target (required), project_id (required), version_id (required), +8 more.
df_list_connectionsBRead-onlyIdempotent
List database connections of a project version. Credentials are never returned. Unsupported engines (e.g. MySQL) are excluded entirely.
Inputs: project_id (required), version_id (required), page, status, +4 more.
df_list_data_martsBRead-onlyIdempotent
List data marts of a project version.
Inputs: project_id (required), version_id (required), page, search, +5 more.
df_list_dimension_groupsARead-onlyIdempotent
List dimension groups (shared reference hierarchies) of a version.
Inputs: project_id (required), version_id (required), page, language, +2 more.
df_list_fact_tablesBRead-onlyIdempotent
List fact tables of a project version with element counts.
Inputs: project_id (required), version_id (required), page, language, +2 more.
df_list_git_connectionsARead-onlyIdempotent
List the company's saved Git connections. Credentials are never returned. Requires a company administrator API key.
Inputs: page, page_size, use_cache.
df_list_projectsBRead-onlyIdempotent
List DataForge projects visible to the configured API key.
Inputs: page, page_size, use_cache.
df_list_relationshipsCRead-onlyIdempotent
List star-schema relationships (fact table to dimension group joins). relationship_type is the raw slug many_to_one.
Inputs: project_id (required), version_id (required), page, language, +4 more.
df_list_versionsBRead-onlyIdempotent
List versions of a DataForge project.
Inputs: project_id (required), page, page_size, use_cache.
df_preview_importAIdempotent
Dry run: compare an import source with the target version and report what would change, including per-field conflicts. Nothing is written.
Inputs: project_id (required), version_id (required), path, branch, +9 more.
df_refresh_cacheAIdempotent
Drop the cached state of a project version (or the whole project) and re-fetch its RMD snapshot.
Inputs: project_id (required), version_id (required), scope, language.
df_remove_group_dimensionADestructiveIdempotent
PERMANENTLY DELETES DATA IN DATAFORGE. Remove one dimension from a group. The dimension itself stays in the RMD.
Inputs: project_id (required), version_id (required), dimension_id (required), dimension_group_id (required).
df_revoke_project_accessADestructiveIdempotent
PERMANENTLY DELETES DATA IN DATAFORGE. Remove a user's explicit access to a project.
Inputs: user_id (required), project_id (required).
df_set_group_dimensionsADestructiveIdempotent
WRITES TO DATAFORGE. Add dimensions to a group, or change the hierarchy level of existing members. Applied all-or-nothing; levels must stay unique.
Inputs: dimensions (required), project_id (required), version_id (required), dimension_group_id (required), +1 more.
df_set_project_accessADestructiveIdempotent
WRITES TO DATAFORGE. Grant or change a user's access level on a project. Downgrade-only: a level above the user's global role is rejected. Requires the caller to be the project owner or a company administrator.
Inputs: user_id (required), project_id (required), access_level (required), idempotency_key.
df_test_git_connectionAIdempotent
WRITES TO DATAFORGE. Run the five repository checks for a saved Git connection and refresh its stored status. A failed check is reported as status=failed, not as an error.
Inputs: connection_id (required).
df_transfer_project_ownershipADestructiveIdempotent
WRITES TO DATAFORGE. Transfer project ownership. The new owner must belong to the project's company and hold an ownership-capable role.
Inputs: project_id (required), new_owner_id (required).
df_unassign_from_fact_tableADestructiveIdempotent
PERMANENTLY DELETES DATA IN DATAFORGE. Detach one element from a fact table. The element itself stays in the RMD.
Inputs: element_id (required), project_id (required), version_id (required), element_type (required), +1 more.
df_update_git_connectionADestructiveIdempotent
WRITES TO DATAFORGE. Replace a saved Git connection's configuration.
Inputs: connection_id (required), name, path, branch, +4 more.
df_update_projectBDestructiveIdempotent
WRITES TO DATAFORGE. Update a project's name, description or colour.
Inputs: project_id (required), name, color, description.
df_update_versionADestructiveIdempotent
WRITES TO DATAFORGE. Rename a version or make it the global (published) one. Versions have no description field.
Inputs: project_id (required), version_id (required), name, is_global.
df_write_dimensionADestructiveIdempotent
WRITES TO DATAFORGE. Create, replace or update a dimension in a project version. mode=create adds a new one; mode=replace (PUT) overwrites it and RESETS every optional field you do not pass; mode=update (PATCH) changes only the fields you pass. replace and update require dimension_id.
Inputs: project_id (required), version_id (required), id, mode, +21 more.
df_write_dimension_groupCDestructiveIdempotent
WRITES TO DATAFORGE. Create, replace or update a dimension group. primary_key is the source object of the group's key column.
Inputs: project_id (required), version_id (required), mode, name, +5 more.
df_write_factADestructiveIdempotent
WRITES TO DATAFORGE. Create, replace or update a fact in a project version. mode=create adds a new one; mode=replace (PUT) overwrites it and RESETS every optional field you do not pass; mode=update (PATCH) changes only the fields you pass. replace and update require fact_id.
Inputs: project_id (required), version_id (required), id, mode, +19 more.
df_write_fact_tableADestructiveIdempotent
WRITES TO DATAFORGE. Create, replace or update a fact table. A fact table created via the API has no base physical table; attach elements with df_assign_to_fact_table.
Inputs: project_id (required), version_id (required), mode, name, +3 more.
df_write_measureADestructiveIdempotent
WRITES TO DATAFORGE. Create, replace or update a measure in a project version. mode=create adds a new one; mode=replace (PUT) overwrites it and RESETS every optional field you do not pass; mode=update (PATCH) changes only the fields you pass. replace and update require measure_id.
Inputs: project_id (required), version_id (required), id, mode, +21 more.
df_write_relationshipBDestructiveIdempotent
WRITES TO DATAFORGE. Create, replace or update a star-schema relationship. foreign_key (fact table side) and primary_key (dimension group side) must name the same connection - a join cannot span two databases.
Inputs: project_id (required), version_id (required), mode, foreign_key, +6 more.
df_write_verification_filterADestructiveIdempotent
WRITES TO DATAFORGE. Create, replace or update a verification filter. Pass fact_table_id to target a fact-table filter; omit it for a version-level (global) filter. Element references in conditions are written as [Element name].
Inputs: project_id (required), version_id (required), mode, name, +5 more.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 65 tools
Most tools are clearly separated by resource type (measure, dimension, fact, fact_table, dimension_group, relationship, version, project, connection, git_connection, data_mart) and action (get/list/write/delete/bulk/import/export). A few potential confusions exist: df_get_rmd vs df_get_consolidated_rmd overlap heavily (one is a normalized subset of the other), and df_get_connection_schema vs df_get_connection with include_db_schema=true could be confused, but descriptions clarify the distinction.
All tools follow a consistent df_<verb>_<resource> pattern with snake_case throughout. Verbs are predictable: get/list for reads, write for create/replace/update, delete for removals, bulk_write for batch operations, and import/export for transfer operations. The prefix df_ is applied uniformly, making the set highly predictable.
65 tools is a very large surface for a single MCP server. While the domain is broad (semantic modeling, fact tables, relationships, versions, projects, git, access control, import/export), the count exceeds the 25+ threshold and will impose significant context overhead on agents. The tools could reasonably be split into multiple focused servers (e.g., semantic model editor, admin/access, git/import-export).
The tool surface is remarkably complete for the DataForge domain: full CRUD for projects, versions, measures, dimensions, facts, fact tables, dimension groups, relationships, verification filters, and git connections; plus bulk operations, assignment/detachment, access management, import/export with dry-run validation, and health/cache utilities. No obvious dead ends or missing lifecycle operations were identified.
Maintenance
Related MCP Connectors
Persistent memory, knowledge graph, task management and more for AI agents via MCP.
Shared memory for connected AI tools. Projects, rules and skills over MCP. OAuth or API key.
Cross-vendor AI memory over MCP. One semantic store, readable and writeable from every MCP client.
Hosted MCP server for live public-data APIs and Skills for AI agents.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceEnables AI agents to query live schema, lineage, and query-context across data warehouses, dbt projects, orchestration systems, and BI tools via MCP tools.Apache 2.0
- FlicenseNot gradedqualityDmaintenanceThis server exposes a Microsoft Fabric Data Agent as a native MCP tool, allowing clients to query the agent for KPIs, financial data, and other quantified project intelligence via natural language.-
- AlicenseNot gradedqualityDmaintenanceMCP server providing read-only Snowflake metadata tools (schemas, tables, queries, lineage) for agentic data pipeline generation, enabling natural-language-to-pipeline workflows with dbt, Airflow, and Great Expectations.1MIT
- AlicenseNot gradedqualityDmaintenanceEnables enterprise AI agents to query governed data lineage, PII-aware schema documentation, and semantic metadata from SQL logs via MCP, with role-based access and vector search.MIT