fix(github-copilot): preserve reasoning IDs for Copilot Codex models (#71684)

* fix(github-copilot): preserve all reasoning IDs and add gpt-5.3-codex support

The existing guard (8fd15ed0e5) only skipped rewriting reasoning item IDs
when encrypted_content was a non-null string. When gpt-5.3-codex is used
via GitHub Copilot, the model falls through to the forward-compat catch-all
with reasoning:false, so encrypted_content is never requested and arrives
as null — bypassing the guard and causing a rewrite. Copilot validates
reasoning item IDs server-side regardless of whether the client includes
encrypted_content, so the rewritten id triggers the 400 error.

Two changes:

1. connection-bound-ids.ts: skip ALL reasoning items unconditionally.
   Reasoning items always reference server-side state bound to their
   original ID; rewriting any of them breaks Copilot's lookup.

2. models.ts + index.ts: extend the forward-compat cloning logic to
   cover gpt-5.3-codex (adds it to the template-target set and to
   CODEX_TEMPLATE_MODEL_IDS so it can also serve as a template source
   for gpt-5.4). Adds gpt-5.3-codex to COPILOT_XHIGH_MODEL_IDS for
   the thinking profile.

Thanks @InvalidPandaa.

* docs(github-copilot): clarify gpt-5.3-codex is a no-op template for itself

https://claude.ai/code/session_01EAFmq4WyKkiUkVAqRXp4Bm

* fix(github-copilot): remove dead reasoning prefix branch in deriveReplacementId

https://claude.ai/code/session_01EAFmq4WyKkiUkVAqRXp4Bm

* fix(github-copilot): align reasoning id replay tests

* test(plugin-sdk): use cjs sidecar for require fast path

---------

Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: Peter Steinberger <steipete@gmail.com>

extensions/github-copilot/models.ts

@@ -6,12 +6,18 @@ import { normalizeModelCompat } from "openclaw/plugin-sdk/provider-model-shared"
 import { normalizeOptionalLowercaseString } from "openclaw/plugin-sdk/text-runtime";
 
 export const PROVIDER_ID = "github-copilot";
-const CODEX_GPT_54_MODEL_ID = "gpt-5.4";
-const CODEX_TEMPLATE_MODEL_IDS = ["gpt-5.2-codex"] as const;
+const CODEX_FORWARD_COMPAT_TARGET_IDS = new Set(["gpt-5.4", "gpt-5.3-codex"]);
+// gpt-5.3-codex is only a useful template when gpt-5.4 is the target; it is
+// always a registry miss (and therefore skipped) when it is the target itself.
+const CODEX_TEMPLATE_MODEL_IDS = ["gpt-5.3-codex", "gpt-5.2-codex"] as const;
 
 const DEFAULT_CONTEXT_WINDOW = 128_000;
 const DEFAULT_MAX_TOKENS = 8192;
 
+function isCopilotCodexModelId(modelId: string): boolean {
+  return /(?:^|[-_.])codex(?:$|[-_.])/.test(modelId);
+}
+
 export function resolveCopilotTransportApi(
   modelId: string,
 ): "anthropic-messages" | "openai-responses" {
@@ -35,9 +41,9 @@ export function resolveCopilotForwardCompatModel(
     return undefined;
   }
 
-  // For gpt-5.4 specifically, clone from the gpt-5.2-codex template
-  // to preserve any special settings the registry has for codex models.
-  if (lowerModelId === CODEX_GPT_54_MODEL_ID) {
+  // For gpt-5.4 and gpt-5.3-codex, clone from a registered codex template
+  // to inherit the correct reasoning and capability flags.
+  if (CODEX_FORWARD_COMPAT_TARGET_IDS.has(lowerModelId)) {
     for (const templateId of CODEX_TEMPLATE_MODEL_IDS) {
       const template = ctx.modelRegistry.find(
         PROVIDER_ID,
@@ -60,7 +66,7 @@ export function resolveCopilotForwardCompatModel(
   // model isn't available on the user's plan. This lets new models be used
   // by simply adding them to agents.defaults.models in openclaw.json — no
   // code change required.
-  const reasoning = /^o[13](\b|$)/.test(lowerModelId);
+  const reasoning = /^o[13](\b|$)/.test(lowerModelId) || isCopilotCodexModelId(lowerModelId);
   return normalizeModelCompat({
     id: trimmedModelId,
     name: trimmedModelId,