[dinwwwh]

Requested by dinwwwh · project thread

Summary

Docs only: this adds one sentence to the Hey API bullet in the downgrader README.

Hey API 0.99 drops a { type: "object" } branch that has nothing else in it from an anyOf, but only when the schema holds a readOnly or writeOnly property somewhere, even a nested one. The drop shows up in its types, in Zod and in Valibot. A 3.1 type: ["object", "string"] becomes that anyOf in 3.0, so Hey's 3.0 Zod and Valibot schemas reject an object there, while the 3.1 input accepts it.

Evidence

• Real docs: I ran the Hey Zod runtime oracle on 83 OpenAPI 3.1/3.2 docs from npm packages published since Nov 1. It flagged 6 values, all in the Celigo/Boomi api-specs set (for example HTTP.encrypted and MCP.http.encrypted), with "expected string, received object".
• Minimal case: the trigger is a readOnly or writeOnly sibling, at any depth, next to { anyOf: [{ type: object }, { type: string }] }. Without such a sibling, Hey keeps the union.
• Not caused by 3.0: Hey drops the branch in the same way when a 3.1 doc writes that anyOf itself.
• Avoiding it: adding additionalProperties: true to the branch keeps the union. Because this is Hey's own pruning, the change is a README note, not a code change.

---

Generated by Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[codspeed]

Merging this PR will not alter performance

✅ 29 untouched benchmarks
⏩ 2 skipped benchmarks¹

---

_{Comparing claude/project-thread-ttohne (2a72f89) with main (d60844d)}

Footnotes

1. 2 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[pullfrog]

✅ No new issues found.

Reviewed changes

Docs-only change adding one sentence to the Hey API limitation bullet in packages/downgrader/README.md.

• Hey API bare-object anyOf branch drop — notes that Hey API drops a { type: "object" } anyOf branch with nothing else in it, which a downgraded 3.1 type: ["object", …] becomes, when the schema holds a readOnly/writeOnly property, so Hey's types, Zod and Valibot then reject an object (Celigo encrypted fields).

The sentence reads consistently with the surrounding bullet (same also cadence, such as examples, "its types and those plugins" referring to the Zod/Valibot plugins named earlier), is scoped to the downgraded 3.0 output that the type-list → anyOf conversion produces (v3.1-to-v3.0.ts:845-869), and does not contradict the bullet's closing "Hey API reads 3.1 too…" claim. uncheck (oxfmt) passes on the file.

  ^{｜ View workflow run ｜ Using deepseek-v4.1-flash (free via Pullfrog for OSS) ｜ 𝕏}