fix: Address PR review feedback - fix stale resume IDs, update model description, remove duplicate test, revert gitignore

Author: hannesrudolph

.gitignore

@@ -49,7 +49,3 @@ logs
 
 # Qdrant
 qdrant_storage/
-
-# ignore temp background docs
-TEMP_OPENAI_BACKGROUND_TASK_DOCS.DM
-TEMP_DOCS/

packages/types/src/providers/openai.ts

@@ -47,7 +47,7 @@ export const openAiNativeModels = {
 		inputPrice: 15.0,
 		outputPrice: 120.0,
 		description:
-			"GPT-5 Pro: a slow, reasoning-focused model built to tackle tough problems. Requests can take several minutes to finish. Responses API only; no streaming, so it may appear stuck until the reply is ready.",
+			"GPT-5 Pro: A slow, reasoning-focused model for complex problems. Uses background mode with resilient streaming - requests may take several minutes with automatic recovery if connection drops.",
 		supportsVerbosity: true,
 		supportsTemperature: false,
 		backgroundMode: true,

src/api/providers/openai-native.ts

@@ -54,6 +54,9 @@ export class OpenAiNativeHandler extends BaseProvider implements SingleCompletio
 	// Track whether current request is in background mode for status chunk annotation
 	private currentRequestIsBackground?: boolean
 	private resumeCutoffSequence?: number
+	// Per-request tracking to prevent stale resume attempts
+	private currentRequestResponseId?: string
+	private currentRequestSequenceNumber?: number
 
 	// Event types handled by the shared event processor to avoid duplication
 	private readonly coreHandledEventTypes = new Set<string>([
@@ -325,12 +328,15 @@ export class OpenAiNativeHandler extends BaseProvider implements SingleCompletio
 	): ApiStream {
 		// Annotate if this request uses background mode (used for status chunks)
 		this.currentRequestIsBackground = !!requestBody?.background
+		// Reset per-request tracking to prevent stale values from previous requests
+		this.currentRequestResponseId = undefined
+		this.currentRequestSequenceNumber = undefined
 
 		const canAttemptResume = () =>
 			this.currentRequestIsBackground &&
 			(this.options.openAiNativeBackgroundAutoResume ?? true) &&
-			!!this.lastResponseId &&
-			typeof this.lastSequenceNumber === "number"
+			!!this.currentRequestResponseId &&
+			typeof this.currentRequestSequenceNumber === "number"
 
 		try {
 			// Use the official SDK
@@ -356,8 +362,8 @@ export class OpenAiNativeHandler extends BaseProvider implements SingleCompletio
 				// Stream dropped mid-flight; attempt resume for background requests
 				if (canAttemptResume()) {
 					for await (const chunk of this.attemptResumeOrPoll(
-						this.lastResponseId!,
-						this.lastSequenceNumber!,
+						this.currentRequestResponseId!,
+						this.currentRequestSequenceNumber!,
 						model,
 					)) {
 						yield chunk
@@ -411,8 +417,8 @@ export class OpenAiNativeHandler extends BaseProvider implements SingleCompletio
 						} catch (fallbackErr) {
 							if (canAttemptResume()) {
 								for await (const chunk of this.attemptResumeOrPoll(
-									this.lastResponseId!,
-									this.lastSequenceNumber!,
+									this.currentRequestResponseId!,
+									this.currentRequestSequenceNumber!,
 									model,
 								)) {
 									yield chunk
@@ -436,8 +442,8 @@ export class OpenAiNativeHandler extends BaseProvider implements SingleCompletio
 						}
 						if (canAttemptResume()) {
 							for await (const chunk of this.attemptResumeOrPoll(
-								this.lastResponseId!,
-								this.lastSequenceNumber!,
+								this.currentRequestResponseId!,
+								this.currentRequestSequenceNumber!,
 								model,
 							)) {
 								yield chunk
@@ -466,8 +472,8 @@ export class OpenAiNativeHandler extends BaseProvider implements SingleCompletio
 						}
 						if (canAttemptResume()) {
 							for await (const chunk of this.attemptResumeOrPoll(
-								this.lastResponseId!,
-								this.lastSequenceNumber!,
+								this.currentRequestResponseId!,
+								this.currentRequestSequenceNumber!,
 								model,
 							)) {
 								yield chunk
@@ -489,8 +495,8 @@ export class OpenAiNativeHandler extends BaseProvider implements SingleCompletio
 				}
 				if (canAttemptResume()) {
 					for await (const chunk of this.attemptResumeOrPoll(
-						this.lastResponseId!,
-						this.lastSequenceNumber!,
+						this.currentRequestResponseId!,
+						this.currentRequestSequenceNumber!,
 						model,
 					)) {
 						yield chunk
@@ -500,8 +506,10 @@ export class OpenAiNativeHandler extends BaseProvider implements SingleCompletio
 				throw fallbackErr
 			}
 		} finally {
-			// Always clear background flag at end of request lifecycle
+			// Always clear background flag and per-request tracking at end of request lifecycle
 			this.currentRequestIsBackground = undefined
+			this.currentRequestResponseId = undefined
+			this.currentRequestSequenceNumber = undefined
 		}
 	}
 
@@ -818,11 +826,15 @@ export class OpenAiNativeHandler extends BaseProvider implements SingleCompletio
 							// Record sequence number for cursor tracking
 							if (typeof parsed?.sequence_number === "number") {
 								this.lastSequenceNumber = parsed.sequence_number
+								// Also track for per-request resume capability
+								this.currentRequestSequenceNumber = parsed.sequence_number
 							}
 
 							// Store response ID for conversation continuity
 							if (parsed.response?.id) {
 								this.resolveResponseId(parsed.response.id)
+								// Also track for per-request resume capability
+								this.currentRequestResponseId = parsed.response.id
 							}
 							// Capture resolved service tier if present
 							if (parsed.response?.service_tier) {
@@ -1493,6 +1505,8 @@ export class OpenAiNativeHandler extends BaseProvider implements SingleCompletio
 		// Persist response id for conversation continuity when available
 		if (event?.response?.id) {
 			this.resolveResponseId(event.response.id)
+			// Also track for per-request resume capability
+			this.currentRequestResponseId = event.response.id
 		}
 		// Capture resolved service tier when available
 		if (event?.response?.service_tier) {
@@ -1501,6 +1515,8 @@ export class OpenAiNativeHandler extends BaseProvider implements SingleCompletio
 		// Record sequence number for cursor tracking
 		if (typeof event?.sequence_number === "number") {
 			this.lastSequenceNumber = event.sequence_number
+			// Also track for per-request resume capability
+			this.currentRequestSequenceNumber = event.sequence_number
 		}
 
 		// Map lifecycle events to status chunks