[josemontespg]

Summary

Adds the transformData data function to the v1 MCP catalog and @a2ui/catalog-mcp. transformData evaluates a standard JMESPath expression, passed in its jmespath argument, against data, letting a surface select, filter, and reshape fields from a callMcpTool payload into the {path: value} object that updateDataModel writes.

transformData is declared in catalogs/mcp/v1/catalog.json, registered on mcpCatalog, and exported as TransformDataApi and TransformDataImplementation. It implements standard JMESPath with no extensions, raises an A2uiExpressionError when an expression fails to parse (Invalid JMESPath expression: ...) or evaluate, and treats @path or @call objects inside data as plain data rather than expressions.

Changes

• typescript/catalogs/mcp/src/functions/transform_data.ts and transform_data.test.ts: implement TransformDataApi and TransformDataImplementation with unit and JMESPath compliance tests.
• typescript/catalogs/mcp/src/catalog.ts and src/index.ts: register TransformDataImplementation on mcpCatalog and export TransformDataApi and TransformDataImplementation.
• catalogs/mcp/v1/catalog.json and typescript/catalogs/mcp/src/catalog.test.ts: declare transformData with links to the JMESPath specification and check parity with mcpCatalog.
• typescript/catalogs/mcp/package.json and yarn.lock: add jmespath and @types/jmespath.
• catalogs/mcp/README.md, typescript/catalogs/mcp/README.md, CHANGELOG.md: document transformData and how updateDataModel(transformData(callMcpTool(...))) chains settle async tool results.

Verification

• yarn workspace @a2ui/catalog-mcp run test and yarn workspace @a2ui/catalog-mcp run lint: node and karma suites pass, including transform_data.test.ts and the callMcpTool.test.ts payload chaining case.
• specification/v1_0/test/run_tests.py and ./scripts/fix_format.sh --check pass.

[gemini-code-assist]

Code Review

This pull request introduces the transformData function to the A2UI MCP catalog, allowing surfaces to evaluate standard JMESPath expressions against tool results to shape them into data model updates. The changes include the implementation of the function, schema updates, comprehensive unit tests, and documentation. The reviewer pointed out that the placeholder [#NNNN] in the changelog should be replaced with the actual pull request number before merging.