Refactor reasoning handling and metadata management in responses

- Updated chat stream tests to clarify reasoning behavior and ensure correct handling of include lists.
- Enhanced responses proxy route to normalize include lists and apply default reasoning includes based on request context.
- Improved transformer metadata handling, ensuring proper serialization and deserialization of metadata records.
- Added tests for preserving reasoning signatures and managing citations in OpenAI-compatible responses.
- Refined the logic for injecting default includes in responses and ensured compatibility with codex surface defaults.

src/server/transformers/openai/chat/helpers.ts

@@ -7,6 +7,7 @@ import type {
   OpenAiChatToolCall,
   OpenAiChatUsageDetails,
 } from './model.js';
+import { fromTransformerMetadataRecord } from '../../shared/normalized.js';
 
 function isRecord(value: unknown): value is Record<string, unknown> {
   return !!value && typeof value === 'object' && !Array.isArray(value);
@@ -345,7 +346,9 @@ function extractCitationsFromPayload(
     citations.add(citation);
   }
 
-  const transformerMetadata = asRecord(payload.transformer_metadata) ?? asRecord(payload.transformerMetadata);
+  const transformerMetadata = fromTransformerMetadataRecord(
+    payload.transformer_metadata ?? payload.transformerMetadata,
+  );
   if (transformerMetadata) {
     for (const citation of dedupeCitations(transformerMetadata.citations)) {
       citations.add(citation);

src/server/transformers/openai/chat/index.test.ts

@@ -1,6 +1,7 @@
 import { describe, expect, it } from 'vitest';
 
 import { createClaudeDownstreamContext } from '../../shared/normalized.js';
+import { toTransformerMetadataRecord } from '../../shared/normalized.js';
 import { openAiChatTransformer } from './index.js';
 
 function parseSsePayloads(lines: string[]): Array<Record<string, unknown>> {
@@ -154,6 +155,32 @@ describe('openAiChatTransformer.outbound', () => {
     });
   });
 
+  it('consumes shared transformer metadata citations on openai-compatible final responses', () => {
+    const normalized = openAiChatTransformer.transformFinalResponse({
+      id: 'chatcmpl-shared-metadata',
+      model: 'gpt-5',
+      created: 123,
+      transformer_metadata: toTransformerMetadataRecord({
+        citations: [{ uri: 'https://shared.example/citation' }],
+        thoughtSignature: 'sig-final',
+        thoughtSignatures: ['sig-tool', 'sig-final'],
+        passthrough: {
+          cachedContent: 'cached/item-1',
+        },
+      }),
+      choices: [{
+        index: 0,
+        finish_reason: 'stop',
+        message: {
+          role: 'assistant',
+          content: 'hello',
+        },
+      }],
+    }, 'gpt-5');
+
+    expect((normalized as any).citations).toEqual(['https://shared.example/citation']);
+  });
+
   it('serializes multi-choice chat responses with per-choice annotations, reasoning, tool calls, and shared citations', () => {
     const normalized = openAiChatTransformer.transformFinalResponse({
       id: 'chatcmpl-multi',