[gonisulaimann]

Problem

1. Missing WebSocket transport on first-class executor: When selecting the first-class codex-app-server provider directly from the executor registry (open-sse/executors/index.ts), the lazy factory instantiated new CodexAppServerExecutor({}, "codex-app-server") with empty options. Without a websocketFn, the executor had no transport available to connect to the app-server.
2. Model reasoning alias not stripped: In CodexAppServerExecutor.execute, input.model was forwarded verbatim to the turn/start JSON-RPC request. Dispatches with suffixed aliases like gpt-5.5-medium or gpt-5.6-sol-ultra failed upstream because the server does not recognize reasoning-suffixed aliases as model names.

Reported in #14277.

Fix

1. In open-sse/executors/index.ts, wire the shared WebSocket transport (codex.getCodexAppServerWebsocketTransport()) into the codex-app-server factory while keeping the lazy loading contract intact.
2. In open-sse/executors/codex-app-server.ts:
   • Import splitCodexReasoningSuffix from ./codex/reasoningSuffix.ts.
   • Forward baseModel as model on the turn/start payload.
   • Forward the reasoning effort separately, giving explicit model suffix aliases precedence over request body defaults (consistent with [BUG] GPT-5.5 Codex suffix aliases are overridden by client default reasoning.effort #2331).
   • Support body.reasoning_effort in extractEffort in addition to body.reasoning.effort.

Verification

Extended tests/unit/codex-app-server.test.ts with 4 new tests:

• turn/start receives baseModel and suffix-derived effort (verifies gpt-5.5-medium -> model: gpt-5.5, effort: medium)
• model suffix effort takes precedence over body effort
• body effort is forwarded when model has no reasoning suffix
• registry loader injects codex websocket transport (verifies executor obtained via getExecutor("codex-app-server") dispatches turns over the shared transport)

All 34 tests in tests/unit/codex-app-server.test.ts pass.
tests/unit/executor-map-golden.test.ts passes.
npm run check:cycles passes (0 cycles).

Closes #14277.

[diegosouzapw]

Confirmed this mirrors the exact working pattern already used in codex.ts's internal fallback path (websocketFn: getCodexAppServerWebsocketTransport()) — the registry factory was genuinely the one place missing it. Good catch on the reasoning-suffix precedence too. CI reds here are the unrelated inherited base-red (issue #13866).

Two small items we'll handle at merge unless you get to them first: a changelog.d fragment (merge-integrity gate), and a Co-authored-by trailer for @aref-alapour, who proposed this exact patch in #14277.

[aref-alapour]

Thanks for the credit and for picking this up @diegosouzapw

[diegosouzapw]

Thanks @gonisulaimann — merging via the release merge-train. Validated in local merge-train (mt-train10c) on the devbox @ train tip 4d841aa1c740bbaa03868dc0a403c62099a99a42 with the 72 sibling PRs of this batch: typecheck:core, file-size, complexity, cognitive-complexity, changelog-integrity green; changed-area node:test 831/831 (0 failing) and vitest 480/482 — the two reds are autoCombo/provider-family-combos.test.ts timing out at 20s, which reproduces on the PURE release tip under the full vitest suite (and is already tracked by the Release-Green issue #13866), so it is inherited, not this batch's. Merged --admin per merge-gates §3/§4/§7.