sdmx-audit-mcp
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., "@sdmx-audit-mcpShow me the codelists available in the registry"
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.
sdmx-audit-mcp
First-party Python MCP server exposing tools to create and update agencies
and codelists in the lab's Fusion Metadata Registry (FMR), backed by
server-side pysdmx. Runs directly with
uv run server.py (PEP 723 inline dependencies, no install step) - all files
are plain sibling modules, not an installed package.
Contents
File | Purpose |
|
|
| Read access via |
| Agency list/create/update logic, including the id-normalisation needed because |
| Codelist list/get/create/update logic (in-place code merge). |
|
|
| Alternate entrypoint ( |
Related MCP server: Shiranui
Why the official mcp SDK instead of fastmcp
This server uses the official
mcp Python SDK's
mcp.server.fastmcp.FastMCP, which supports Streamable HTTP transport
without an extra third-party dependency. Pinned to mcp<2: the installed-by-
default mcp 2.x renamed FastMCP to MCPServer with a different API.
Agency id normalisation
RegistryClient.get_agencies(parent_agency) returns each Agency.id fully
qualified with its ancestor path (e.g. BIS.CBS for the CBS agency under
BIS), even though the underlying SDMX-ML stores just the short local id
(CBS). Resubmitting an AgencyScheme with the fully-qualified id fails
FMR's id-format validation (dots are not a valid SDMX id character). See
agencies._short_id() - all create/update logic normalises to the short id
before constructing artefacts for submission, and accepts either form as
input.
In-place merge safety
create_agency/update_agency/update_codelist all fetch the current
artefact, merge the change in Python, and resubmit the complete artefact
(entire agency list, or entire code list). This was empirically verified
against the lab's local FMR: resubmitting the same Codelist id+version with
an expanded code list correctly results in the union of codes, not a
destructive replace - see the package tests.
Local build & run
# Runs directly - dependencies are declared inline in server.py (PEP 723),
# uv installs them into an ephemeral environment on first run.
FMR_SDMX_URL=http://localhost:5000/sdmx/v2 \
FMR_SUBMIT_URL=http://localhost:5000/ws/secure/sdmxapi/rest/ \
FMR_ROOT_USER=demo \
FMR_ROOT_PASSWORD=password \
uv run server.pyInside the aio-sandbox, fmr (the in-network service name) resolves on the
shared docker network; standalone runs need the labspace stack up and FMR
reachable at the URLs above (adjust host/port to match).
Tests
uv run pytestTests call the tool-implementation functions (agencies.py/codelists.py)
directly against a live FMR instance (skipped if unreachable), rather than
driving the MCP JSON-RPC transport - the FastMCP tool-dispatch wrapper in
server.py is trivial and adds no behaviour worth re-testing through the
protocol layer.
This server cannot be deployed
Maintenance
Related MCP Connectors
Query, browse, and automate OmegaAI workspaces from any MCP client. Streamable HTTP with OAuth 2.0.
Governed data discovery, exact queries, decisions, simulations, and runtime utilities over MCP.
Public read-only MCP for products, frameworks, guides, methodology, and blog metadata.
- n3tz AtlasOAuthai.n3tz
Governed company knowledge over MCP. Access by contract for authorized tenants.
Related MCP Servers
- AlicenseAqualityCmaintenanceEnables querying SAP SuccessFactors OData API metadata and managing Role-Based Permission (RBP) configurations. It provides tools for retrieving entity metadata, listing permission roles, and inspecting user-specific access rights through MCP-compatible clients.2911MIT
- AlicenseCqualityCmaintenanceMCP server for retrieving clinical standard content from the CDISC Library, supporting controlled terminology, ADaM, SDTM, CDASH, SEND metadata, and search.317MIT
- FlicenseNot gradedqualityCmaintenanceEnables MCP-compliant agents to perform data discovery, schema matching, and export via HTTP, with a mock mode for demonstration.-
- AlicenseNot gradedqualityAmaintenanceEnables MCP clients to serve and query semantic models, providing tools for entity descriptions, metric lookups, context resolution, and operation validation for AI agents.MIT