fix(openai-codex): address Luna Responses Lite review feedback

- Capture telemetry when transformLunaResponsesLiteBody throws in createMessage
- Drop weak Record<string, any> alias from transformLunaResponsesLiteBody signature
- Document intentional tool_choice/parallel_tool_calls and reasoning context overrides
- Strengthen tests: full input array assertion, reasoning context overwrite case,
  retry header coverage, single-build spy, and uuidv7 session shape check

Author: roomote

src/api/providers/__tests__/openai-codex.spec.ts

@@ -275,6 +275,19 @@ describe("transformLunaResponsesLiteBody", () => {
 		})
 	})
 
+	it("overwrites a pre-existing reasoning context with all_turns", () => {
+		const result = transformLunaResponsesLiteBody(
+			{
+				model: "gpt-5.6-luna",
+				input: [{ role: "user", content: [{ type: "input_text", text: "Hello" }] }],
+				reasoning: { effort: "high", context: "current_turn" },
+			},
+			"session-1",
+		)
+
+		expect(result.reasoning).toEqual({ effort: "high", context: "all_turns" })
+	})
+
 	it.each([
 		["input", { input: "invalid" }, "input must be an array"],
 		["tools", { input: [], tools: {} }, "tools must be an array when provided"],
@@ -325,12 +338,17 @@ describe("OpenAiCodexHandler Luna Responses Lite requests", () => {
 		})
 		expect(body).not.toHaveProperty("tools")
 		expect(body).not.toHaveProperty("instructions")
+		expect(body.input).toHaveLength(3)
 		expect(body.input[0]).toMatchObject({ type: "additional_tools", role: "developer" })
 		expect(body.input[1]).toEqual({
 			type: "message",
 			role: "developer",
 			content: [{ type: "input_text", text: "Luna instructions" }],
 		})
+		expect(body.input[2]).toEqual({
+			role: "user",
+			content: [{ type: "input_text", text: "Hello" }],
+		})
 		expect(options.headers).toMatchObject({
 			originator: "zoo-code",
 			session_id: "task-luna",
@@ -389,6 +407,7 @@ describe("OpenAiCodexHandler Luna Responses Lite requests", () => {
 
 	it("preserves Luna session affinity while retrying with refreshed authentication", async () => {
 		const handler = new OpenAiCodexHandler({ apiModelId: "gpt-5.6-luna" })
+		const transformSpy = vitest.spyOn(handler as any, "buildLunaRequestBody")
 		vitest.spyOn(openAiCodexOAuthManager, "getAccessToken").mockResolvedValue("expired-token")
 		vitest.spyOn(openAiCodexOAuthManager, "forceRefreshAccessToken").mockResolvedValue("refreshed-token")
 		vitest.spyOn(openAiCodexOAuthManager, "getAccountId").mockResolvedValue("acct_test")
@@ -432,16 +451,21 @@ describe("OpenAiCodexHandler Luna Responses Lite requests", () => {
 		const retryBody = JSON.parse(retryOptions.body)
 
 		expect(retryBody).toEqual(firstBody)
+		expect(transformSpy).toHaveBeenCalledTimes(1)
 		expect(firstBody.prompt_cache_key).toBe("task-retry")
 		expect(firstOptions.headers).toMatchObject({
 			Authorization: "Bearer expired-token",
 			"session-id": "task-retry",
 			"x-session-affinity": "task-retry",
+			version: "0.144.0",
+			"x-openai-internal-codex-responses-lite": "true",
 		})
 		expect(retryOptions.headers).toMatchObject({
 			Authorization: "Bearer refreshed-token",
 			"session-id": "task-retry",
 			"x-session-affinity": "task-retry",
+			version: "0.144.0",
+			"x-openai-internal-codex-responses-lite": "true",
 		})
 	})
 
@@ -499,7 +523,7 @@ describe("OpenAiCodexHandler Luna Responses Lite requests", () => {
 		const fetchOptions = mockFetch.mock.calls[0][1]
 		const body = JSON.parse(fetchOptions.body)
 		const sessionId = body.prompt_cache_key
-		expect(sessionId).toEqual(expect.any(String))
+		expect(sessionId).toMatch(/^[0-9a-f]{8}-[0-9a-f]{4}-7[0-9a-f]{3}-[89ab][0-9a-f]{3}-[0-9a-f]{12}$/i)
 		expect(body).toMatchObject({
 			model: "gpt-5.6-luna",
 			stream: false,

src/api/providers/openai-codex.ts

@@ -37,8 +37,6 @@ const CODEX_API_BASE_URL = "https://chatgpt.com/backend-api/codex"
 const LUNA_MODEL_ID = "gpt-5.6-luna"
 const LUNA_CODEX_VERSION = "0.144.0"
 
-type ResponsesRequestBody = Record<string, any>
-
 function stripInputImageDetail(value: any): any {
 	if (Array.isArray(value)) {
 		return value.map(stripInputImageDetail)
@@ -55,10 +53,7 @@ function stripInputImageDetail(value: any): any {
 	)
 }
 
-export function transformLunaResponsesLiteBody(
-	requestBody: ResponsesRequestBody,
-	effectiveSessionId: string,
-): ResponsesRequestBody {
+export function transformLunaResponsesLiteBody(requestBody: any, effectiveSessionId: string): any {
 	if (!Array.isArray(requestBody.input)) {
 		throw new Error("Invalid gpt-5.6-luna Responses Lite request: input must be an array.")
 	}
@@ -91,9 +86,13 @@ export function transformLunaResponsesLiteBody(
 				: []),
 			...transformedInput,
 		],
+		// Luna Responses Lite requires these exact values, so they intentionally
+		// override any caller-supplied tool_choice or parallel_tool_calls.
 		tool_choice: "auto",
 		parallel_tool_calls: false,
 		prompt_cache_key: effectiveSessionId,
+		// Luna Responses Lite requires reasoning context "all_turns"; this intentionally
+		// overwrites any context value already present in the incoming reasoning config.
 		reasoning: { ...reasoning, context: "all_turns" },
 	}
 }
@@ -248,10 +247,19 @@ export class OpenAiCodexHandler extends BaseProvider implements SingleCompletion
 		// Notably: max_output_tokens and prompt_cache_retention may be rejected
 		const effectiveSessionId = metadata?.taskId || this.sessionId
 		const baseRequestBody = this.buildRequestBody(model, formattedInput, systemPrompt, reasoningEffort, metadata)
-		const requestBody =
-			model.id === LUNA_MODEL_ID
-				? transformLunaResponsesLiteBody(baseRequestBody, effectiveSessionId)
-				: baseRequestBody
+		let requestBody: any
+		try {
+			requestBody =
+				model.id === LUNA_MODEL_ID
+					? this.buildLunaRequestBody(baseRequestBody, effectiveSessionId)
+					: baseRequestBody
+		} catch (error) {
+			const message = error instanceof Error ? error.message : String(error)
+			TelemetryService.instance.captureException(
+				new ApiProviderError(message, this.providerName, model.id, "createMessage"),
+			)
+			throw error
+		}
 
 		// Make the request with retry on auth failure
 		for (let attempt = 0; attempt < 2; attempt++) {
@@ -281,6 +289,10 @@ export class OpenAiCodexHandler extends BaseProvider implements SingleCompletion
 		}
 	}
 
+	private buildLunaRequestBody(baseRequestBody: any, effectiveSessionId: string): any {
+		return transformLunaResponsesLiteBody(baseRequestBody, effectiveSessionId)
+	}
+
 	private buildRequestBody(
 		model: OpenAiCodexModel,
 		formattedInput: any,
@@ -1272,7 +1284,7 @@ export class OpenAiCodexHandler extends BaseProvider implements SingleCompletion
 
 			const requestBody =
 				model.id === LUNA_MODEL_ID
-					? transformLunaResponsesLiteBody(baseRequestBody, this.sessionId)
+					? this.buildLunaRequestBody(baseRequestBody, this.sessionId)
 					: baseRequestBody
 
 			const url = `${CODEX_API_BASE_URL}/responses`