[HouMinXi]

Summary

• An operator could not turn universal handoff off for a single combo. The create and update schemas omitted universal_handoff, so Zod stripped the field before the route wrote the combo.
• The runtime already reads combo.universal_handoff or combo.universalHandoff and falls back to the process default when neither is stored. The default stays enabled.
• Both keys are now accepted, and a payload that carries only the handoff setting counts as a real update instead of being rejected as empty.

Related Issues

• Closes fix(api): universal_handoff defaults on and cannot be disabled via combo API (schema strips field) #15251

Validation

• Change type: other (combo API schema)
• Focused tests: node --import tsx/esm --test tests/unit/combo-universal-handoff-15251.test.ts (4 pass)
• eslint on the changed file, with the repo suppression file
• Reconciled with release/v3.8.52 @ 23a1148486
• New test covers the change
• Defect injection: removing the schema fields turns all 4 tests red, and the change was restored

Tests Added Or Updated

• tests/unit/combo-universal-handoff-15251.test.ts (added)

Coverage Notes

• The test parses a combo payload that disables handoff and asserts the field survives both the create and the update schema, including an update that carries only that field.

Reviewer Notes

• The process-wide default in open-sse/services/contextHandoff.ts is left at enabled: true. Flipping it would change behavior for every existing combo and is a separate decision.
• When a payload sets both universal_handoff and universalHandoff, the runtime reads the snake_case value first (comboSetup.ts).

[diegosouzapw]

Thanks for the per-combo universal handoff setting. #15459 is the version that landed on release/v3.8.52 (274ea10): it persists and clears the camelCase alias on the combo route and schema, covered by tests/unit/combo-universal-handoff-15251.test.ts.

This PR touches the same schema and test. Closing it so the two handoff changes are not both applied. None of this diff is being reused on top of #15459. The branch stays in place.