Sdcvalidator

What the sdcvalidator MCP server does

The sdcvalidator MCP server gives MCP-capable agents access to SDC4 validation tasks through a local stdio process. It can validate an XML instance against an XSD schema, produce a more detailed validation report, and inspect an XSD for SDC4 schema compliance. The server is intended for SDC4 data models that use restriction-based derivation rather than extension.

Validation results distinguish structural failures from semantic failures. Structural issues include problems such as unknown elements, invalid nesting, and cardinality violations. Semantic issues include type errors, pattern violations, and enumeration mismatches. This distinction lets a caller determine whether an instance should be rejected or reported for further attention.

How it works

The sdcvalidator MCP server communicates using JSON-RPC 2.0 over standard input and output. It does not require an external MCP SDK. The validation implementation is a thin wrapper around the Python xmlschema package and loads the supplied XSD before checking the XML instance.

Schema loading uses strict validation by default. In this mode, invalid restriction derivations can raise an XMLSchemaParseError, including cases where derived element names or types do not agree with the base definition. The underlying Python API also supports lax mode, which collects schema derivation problems, and skip mode, which bypasses schema-level validation. These modes are described for the library API; the MCP tool descriptions specify schema and instance paths plus an optional compliance-check flag.

Setup and configuration

Install the package from PyPI with pip, or install a local source checkout in editable mode. Start the MCP process with sdcvalidator-mcp serve --mcp. The process is local and uses stdio, so an MCP client must launch it and exchange requests through its standard streams.

The validation tools accept filesystem paths rather than embedding XML or XSD content in the request. validate_instance and validate_and_report take schema_path and instance_path. Both also accept check_compliance, which defaults to true. check_schema_compliance takes only schema_path.

Tools and capabilities

• validate_instance returns whether validation passed, an error count, and classified errors.
• validate_and_report returns a detailed report organized into structural and semantic error categories.
• check_schema_compliance checks that an XSD follows the SDC4 restriction-only rule and does not use xsd:extension.

The package also provides command-line programs for direct use outside MCP: sdcvalidate validates XML, sdcvalidator-xml2json converts XML to schema-aware JSON, and sdcvalidator-json2xml converts JSON back to XML using a schema. The Python API exposes the validator, error tiers, compliance-checking functions, and XML/JSON conversion helpers.

Limitations and notes

This server validates against supplied local schema and instance paths; the provided material does not describe remote schema retrieval, hosted deployment, authentication, or client-specific configuration. Schema compliance checking is separate from instance validation: check_schema_compliance examines the XSD and does not validate an XML instance.

The default strict mode may reject schemas with pre-existing derivation problems during loading. Lax and skip validation modes are available through the Python API, but the MCP tool interface described here does not list a validation-mode parameter. SDC4 compliance checking can be disabled for the two instance-validation tools by setting check_compliance to false.