Add Get and Update Test Execution Steps Tools to Zephyr

[sudheshbmsce]

Goal

Add two missing MCP tools for Zephyr test execution steps, enabling AI agents to read and update individual step results during a test execution run.

Design

The Zephyr Scale API already had two untapped endpoints — GET /testexecutions/{id}/teststeps and PUT /testexecutions/{id}/teststeps — with full Zod schemas already defined in rest-api-schemas.ts. Two new tool classes were implemented following the exact same pattern as all other Zephyr tools (GetTestExecution, UpdateTestExecution, etc.) and registered in ZephyrClient.

A schema fix was also applied to GetTestCaseTestSteps200Response: the testData, reflectRef, and testCase fields were changed from .optional() to .nullish() because the real Zephyr API returns null (not undefined) for these fields, causing output validation failures.

Changeset

• New: src/zephyr/tool/test-execution/get-test-execution-steps.ts — zephyr_get_test_execution_steps tool (read-only, paginated, supports testDataRowNumber filter)
• New: src/zephyr/tool/test-execution/update-test-execution-steps.ts — zephyr_update_test_execution_steps tool (idempotent PUT, records actualResult and statusName per step)
• Modified: src/zephyr/client.ts — registered both new tools
• Fixed: src/zephyr/common/rest-api-schemas.ts — GetTestCaseTestSteps200Response testData/reflectRef/testCase accept null
• New: Unit tests for both tools (14 tests)

Testing

• 14 unit tests added and passing (npm run test:run)
• Validated live against Zephyr Scale API (test case QE-T33108, 10 steps returned correctly after schema fix)