@oh-my-pi/pi-ai 12.11.0 → 12.11.2

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@oh-my-pi/pi-ai",
-	"version": "12.11.0",
+	"version": "12.11.2",
 	"description": "Unified LLM API with automatic model discovery and provider configuration",
 	"type": "module",
 	"main": "./src/index.ts",
@@ -63,7 +63,7 @@
 		"@connectrpc/connect-node": "^2.1.1",
 		"@google/genai": "^1.41.0",
 		"@mistralai/mistralai": "^1.14.0",
-		"@oh-my-pi/pi-utils": "12.11.0",
+		"@oh-my-pi/pi-utils": "12.11.2",
 		"@sinclair/typebox": "^0.34.48",
 		"@smithy/node-http-handler": "^4.4.10",
 		"ajv": "^8.18.0",

package/src/models.json

@@ -8593,8 +8593,8 @@
 				"text"
 			],
 			"cost": {
-				"input": 0.3,
-				"output": 1.2,
+				"input": 0.32,
+				"output": 0.8899999999999999,
 				"cacheRead": 0.15,
 				"cacheWrite": 0
 			},
@@ -9834,12 +9834,12 @@
 				"text"
 			],
 			"cost": {
-				"input": 0.39999999999999997,
-				"output": 1.75,
-				"cacheRead": 0.19999999999999998,
+				"input": 0.47,
+				"output": 2,
+				"cacheRead": 0.14100000000000001,
 				"cacheWrite": 0
 			},
-			"contextWindow": 262144,
+			"contextWindow": 131072,
 			"maxTokens": 65535
 		},
 		"moonshotai/kimi-k2.5": {
@@ -17102,7 +17102,7 @@
 	"cursor": {
 		"claude-4.5-opus-high": {
 			"id": "claude-4.5-opus-high",
-			"name": "Claude 4.5 Opus (Cursor)",
+			"name": "Claude 4.5 Opus",
 			"api": "cursor-agent",
 			"provider": "cursor",
 			"baseUrl": "https://api2.cursor.sh",
@@ -17122,7 +17122,7 @@
 		},
 		"claude-4.5-opus-high-thinking": {
 			"id": "claude-4.5-opus-high-thinking",
-			"name": "Claude 4.5 Opus Thinking (Cursor)",
+			"name": "Claude 4.5 Opus (Thinking)",
 			"api": "cursor-agent",
 			"provider": "cursor",
 			"baseUrl": "https://api2.cursor.sh",
@@ -17142,7 +17142,7 @@
 		},
 		"claude-4.5-sonnet": {
 			"id": "claude-4.5-sonnet",
-			"name": "Claude 4.5 Sonnet (Cursor)",
+			"name": "Claude 4.5 Sonnet",
 			"api": "cursor-agent",
 			"provider": "cursor",
 			"baseUrl": "https://api2.cursor.sh",
@@ -17162,7 +17162,7 @@
 		},
 		"claude-4.5-sonnet-thinking": {
 			"id": "claude-4.5-sonnet-thinking",
-			"name": "Claude 4.5 Sonnet Thinking (Cursor)",
+			"name": "Claude 4.5 Sonnet (Thinking)",
 			"api": "cursor-agent",
 			"provider": "cursor",
 			"baseUrl": "https://api2.cursor.sh",
@@ -17180,9 +17180,104 @@
 			"contextWindow": 200000,
 			"maxTokens": 64000
 		},
+		"claude-4.6-opus-high": {
+			"id": "claude-4.6-opus-high",
+			"name": "Claude 4.6 Opus",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 200000,
+			"maxTokens": 64000
+		},
+		"claude-4.6-opus-high-thinking": {
+			"id": "claude-4.6-opus-high-thinking",
+			"name": "Claude 4.6 Opus (Thinking)",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 200000,
+			"maxTokens": 64000
+		},
+		"claude-4.6-sonnet-medium": {
+			"id": "claude-4.6-sonnet-medium",
+			"name": "Claude 4.6 Sonnet",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 200000,
+			"maxTokens": 64000
+		},
+		"claude-4.6-sonnet-medium-thinking": {
+			"id": "claude-4.6-sonnet-medium-thinking",
+			"name": "Claude 4.6 Sonnet (Thinking)",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 200000,
+			"maxTokens": 64000
+		},
 		"composer-1": {
 			"id": "composer-1",
-			"name": "Composer 1 (Cursor)",
+			"name": "Composer 1",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 200000,
+			"maxTokens": 64000
+		},
+		"composer-1.5": {
+			"id": "composer-1.5",
+			"name": "Composer 1.5",
 			"api": "cursor-agent",
 			"provider": "cursor",
 			"baseUrl": "https://api2.cursor.sh",
@@ -17201,7 +17296,7 @@
 		},
 		"default": {
 			"id": "default",
-			"name": "Auto (Cursor)",
+			"name": "Auto",
 			"api": "cursor-agent",
 			"provider": "cursor",
 			"baseUrl": "https://api2.cursor.sh",
@@ -17221,7 +17316,7 @@
 		},
 		"gemini-3-flash": {
 			"id": "gemini-3-flash",
-			"name": "Gemini 3 Flash (Cursor)",
+			"name": "Gemini 3 Flash",
 			"api": "cursor-agent",
 			"provider": "cursor",
 			"baseUrl": "https://api2.cursor.sh",
@@ -17241,7 +17336,7 @@
 		},
 		"gemini-3-pro": {
 			"id": "gemini-3-pro",
-			"name": "Gemini 3 Pro (Cursor)",
+			"name": "Gemini 3 Pro",
 			"api": "cursor-agent",
 			"provider": "cursor",
 			"baseUrl": "https://api2.cursor.sh",
@@ -17261,7 +17356,7 @@
 		},
 		"gpt-5.1-codex-max": {
 			"id": "gpt-5.1-codex-max",
-			"name": "GPT-5.1 Codex Max (Cursor)",
+			"name": "GPT-5.1 Codex Max",
 			"api": "cursor-agent",
 			"provider": "cursor",
 			"baseUrl": "https://api2.cursor.sh",
@@ -17281,7 +17376,7 @@
 		},
 		"gpt-5.1-codex-max-high": {
 			"id": "gpt-5.1-codex-max-high",
-			"name": "GPT-5.1 Codex Max High (Cursor)",
+			"name": "GPT-5.1 Codex Max High",
 			"api": "cursor-agent",
 			"provider": "cursor",
 			"baseUrl": "https://api2.cursor.sh",
@@ -17299,9 +17394,47 @@
 			"contextWindow": 272000,
 			"maxTokens": 128000
 		},
+		"gpt-5.1-codex-mini": {
+			"id": "gpt-5.1-codex-mini",
+			"name": "GPT-5.1 Codex Mini",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 272000,
+			"maxTokens": 64000
+		},
+		"gpt-5.1-high": {
+			"id": "gpt-5.1-high",
+			"name": "GPT-5.1 High",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 200000,
+			"maxTokens": 64000
+		},
 		"gpt-5.2": {
 			"id": "gpt-5.2",
-			"name": "GPT-5.2 (Cursor)",
+			"name": "GPT-5.2",
 			"api": "cursor-agent",
 			"provider": "cursor",
 			"baseUrl": "https://api2.cursor.sh",
@@ -17319,9 +17452,161 @@
 			"contextWindow": 400000,
 			"maxTokens": 128000
 		},
+		"gpt-5.2-codex": {
+			"id": "gpt-5.2-codex",
+			"name": "GPT-5.2 Codex",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 272000,
+			"maxTokens": 64000
+		},
+		"gpt-5.2-codex-fast": {
+			"id": "gpt-5.2-codex-fast",
+			"name": "GPT-5.2 Codex Fast",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 272000,
+			"maxTokens": 64000
+		},
+		"gpt-5.2-codex-high": {
+			"id": "gpt-5.2-codex-high",
+			"name": "GPT-5.2 Codex High",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 272000,
+			"maxTokens": 64000
+		},
+		"gpt-5.2-codex-high-fast": {
+			"id": "gpt-5.2-codex-high-fast",
+			"name": "GPT-5.2 Codex High Fast",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 272000,
+			"maxTokens": 64000
+		},
+		"gpt-5.2-codex-low": {
+			"id": "gpt-5.2-codex-low",
+			"name": "GPT-5.2 Codex Low",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 272000,
+			"maxTokens": 64000
+		},
+		"gpt-5.2-codex-low-fast": {
+			"id": "gpt-5.2-codex-low-fast",
+			"name": "GPT-5.2 Codex Low Fast",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 272000,
+			"maxTokens": 64000
+		},
+		"gpt-5.2-codex-xhigh": {
+			"id": "gpt-5.2-codex-xhigh",
+			"name": "GPT-5.2 Codex Extra High",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 272000,
+			"maxTokens": 64000
+		},
+		"gpt-5.2-codex-xhigh-fast": {
+			"id": "gpt-5.2-codex-xhigh-fast",
+			"name": "GPT-5.2 Codex Extra High Fast",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 272000,
+			"maxTokens": 64000
+		},
 		"gpt-5.2-high": {
 			"id": "gpt-5.2-high",
-			"name": "GPT-5.2 High (Cursor)",
+			"name": "GPT-5.2 High",
 			"api": "cursor-agent",
 			"provider": "cursor",
 			"baseUrl": "https://api2.cursor.sh",
@@ -17339,9 +17624,161 @@
 			"contextWindow": 400000,
 			"maxTokens": 128000
 		},
+		"gpt-5.3-codex": {
+			"id": "gpt-5.3-codex",
+			"name": "GPT-5.3 Codex",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 272000,
+			"maxTokens": 64000
+		},
+		"gpt-5.3-codex-fast": {
+			"id": "gpt-5.3-codex-fast",
+			"name": "GPT-5.3 Codex Fast",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 272000,
+			"maxTokens": 64000
+		},
+		"gpt-5.3-codex-high": {
+			"id": "gpt-5.3-codex-high",
+			"name": "GPT-5.3 Codex High",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 272000,
+			"maxTokens": 64000
+		},
+		"gpt-5.3-codex-high-fast": {
+			"id": "gpt-5.3-codex-high-fast",
+			"name": "GPT-5.3 Codex High Fast",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 272000,
+			"maxTokens": 64000
+		},
+		"gpt-5.3-codex-low": {
+			"id": "gpt-5.3-codex-low",
+			"name": "GPT-5.3 Codex Low",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 272000,
+			"maxTokens": 64000
+		},
+		"gpt-5.3-codex-low-fast": {
+			"id": "gpt-5.3-codex-low-fast",
+			"name": "GPT-5.3 Codex Low Fast",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 272000,
+			"maxTokens": 64000
+		},
+		"gpt-5.3-codex-xhigh": {
+			"id": "gpt-5.3-codex-xhigh",
+			"name": "GPT-5.3 Codex Extra High",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 272000,
+			"maxTokens": 64000
+		},
+		"gpt-5.3-codex-xhigh-fast": {
+			"id": "gpt-5.3-codex-xhigh-fast",
+			"name": "GPT-5.3 Codex Extra High Fast",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 272000,
+			"maxTokens": 64000
+		},
 		"grok-code-fast-1": {
 			"id": "grok-code-fast-1",
-			"name": "Grok (Cursor)",
+			"name": "Grok",
 			"api": "cursor-agent",
 			"provider": "cursor",
 			"baseUrl": "https://api2.cursor.sh",

package/src/providers/openai-responses.ts

@@ -72,7 +72,7 @@ export interface OpenAIResponsesOptions extends StreamOptions {
 	toolChoice?: ToolChoice;
 	/**
 	 * Enforce strict tool call/result pairing when building Responses API inputs.
-	 * Azure OpenAI Responses API requires tool results to have a matching tool call.
+	 * Azure OpenAI and GitHub Copilot Responses paths require tool results to match prior tool calls.
 	 */
 	strictResponsesPairing?: boolean;
 }
@@ -402,7 +402,9 @@ function createClient(
 }
 
 function buildParams(model: Model<"openai-responses">, context: Context, options?: OpenAIResponsesOptions) {
-	const strictResponsesPairing = options?.strictResponsesPairing ?? isAzureOpenAIBaseUrl(model.baseUrl ?? "");
+	const strictResponsesPairing =
+		options?.strictResponsesPairing ??
+		(isAzureOpenAIBaseUrl(model.baseUrl ?? "") || model.provider === "github-copilot");
 	const messages = convertMessages(model, context, strictResponsesPairing);
 
 	const cacheRetention = resolveCacheRetention(options?.cacheRetention);

package/src/utils/discovery/cursor.ts

@@ -1,3 +1,4 @@
+import * as http2 from "node:http2";
 import { create, fromBinary, toBinary } from "@bufbuild/protobuf";
 import { z } from "zod";
 import { getBundledModels } from "../../models";
@@ -5,7 +6,7 @@ import { GetUsableModelsRequestSchema, GetUsableModelsResponseSchema } from "../
 import type { Model } from "../../types";
 
 const CURSOR_DEFAULT_BASE_URL = "https://api2.cursor.sh";
-const CURSOR_DEFAULT_CLIENT_VERSION = "cli-2026.01.09-231024f";
+const CURSOR_DEFAULT_CLIENT_VERSION = "cli-2026.02.13-41ac335";
 const CURSOR_GET_USABLE_MODELS_PATH = "/agent.v1.AgentService/GetUsableModels";
 
 const DEFAULT_CONTEXT_WINDOW = 200_000;
@@ -47,8 +48,6 @@ export interface CursorModelDiscoveryOptions {
 	timeoutMs?: number;
 	/** Optional list of custom Cursor model ids to include in request context. */
 	customModelIds?: string[];
-	/** Optional fetch implementation override for tests. */
-	fetchImpl?: typeof fetch;
 }
 
 /**
@@ -60,34 +59,19 @@ export interface CursorModelDiscoveryOptions {
 export async function fetchCursorUsableModels(
 	options: CursorModelDiscoveryOptions,
 ): Promise<Model<"cursor-agent">[] | null> {
-	const fetchImpl = options.fetchImpl ?? fetch;
-	const timeoutMs = options.timeoutMs ?? 15_000;
-	const signal = AbortSignal.timeout(timeoutMs);
-
+	const timeoutMs = options.timeoutMs ?? 5_000;
 	try {
 		const requestPayload = create(GetUsableModelsRequestSchema, {
 			customModelIds: normalizeCustomModelIds(options.customModelIds),
 		});
-		const response = await fetchImpl(buildCursorUrl(options.baseUrl), {
-			method: "POST",
-			headers: {
-				"content-type": "application/connect+proto",
-				"connect-protocol-version": "1",
-				te: "trailers",
-				authorization: `Bearer ${options.apiKey}`,
-				"x-ghost-mode": "true",
-				"x-cursor-client-version": options.clientVersion ?? CURSOR_DEFAULT_CLIENT_VERSION,
-				"x-cursor-client-type": "cli",
-			},
-			body: encodeConnectUnaryMessage(toBinary(GetUsableModelsRequestSchema, requestPayload)),
-			signal,
-		});
+		const body = toBinary(GetUsableModelsRequestSchema, requestPayload);
+		const baseUrl = (options.baseUrl ?? CURSOR_DEFAULT_BASE_URL).replace(/\/+$/, "");
 
-		if (!response.ok) {
+		const responseBuffer = await fetchViaHttp2(baseUrl, body, options, timeoutMs);
+
+		if (!responseBuffer) {
 			return null;
 		}
-
-		const responseBuffer = new Uint8Array(await response.arrayBuffer());
 		const decoded = decodeGetUsableModelsResponse(responseBuffer);
 		const parsedDecoded = CursorDecodedResponseSchema.safeParse(decoded);
 		if (!parsedDecoded.success) {
@@ -95,15 +79,76 @@ export async function fetchCursorUsableModels(
 		}
 
 		const references = createCursorReferenceMap();
-
 		return normalizeCursorModels(parsedDecoded.data.models, options.baseUrl, references);
 	} catch {
 		return null;
 	}
 }
 
-function buildCursorUrl(baseUrl?: string): string {
-	return `${(baseUrl ?? CURSOR_DEFAULT_BASE_URL).replace(/\/+$/, "")}${CURSOR_GET_USABLE_MODELS_PATH}`;
+function buildRequestHeaders(options: CursorModelDiscoveryOptions): Record<string, string> {
+	return {
+		"content-type": "application/proto",
+		te: "trailers",
+		authorization: `Bearer ${options.apiKey}`,
+		"x-ghost-mode": "true",
+		"x-cursor-client-version": options.clientVersion ?? CURSOR_DEFAULT_CLIENT_VERSION,
+		"x-cursor-client-type": "cli",
+	};
+}
+
+/** HTTP/2 transport required by Cursor API (HTTP/1.1 is rejected with 464). */
+async function fetchViaHttp2(
+	baseUrl: string,
+	body: Uint8Array,
+	options: CursorModelDiscoveryOptions,
+	timeoutMs: number,
+): Promise<Uint8Array | null> {
+	const { promise, resolve } = Promise.withResolvers<Uint8Array | null>();
+	const client = http2.connect(baseUrl);
+	const timer = setTimeout(() => {
+		client.destroy();
+		resolve(null);
+	}, timeoutMs);
+
+	client.on("error", () => {
+		clearTimeout(timer);
+		resolve(null);
+	});
+
+	const req = client.request({
+		":method": "POST",
+		":path": CURSOR_GET_USABLE_MODELS_PATH,
+		...buildRequestHeaders(options),
+	});
+
+	const chunks: Buffer[] = [];
+	req.on("data", (chunk: Buffer) => chunks.push(chunk));
+	req.on("end", () => {
+		clearTimeout(timer);
+		client.close();
+		resolve(new Uint8Array(Buffer.concat(chunks)));
+	});
+	req.on("error", () => {
+		clearTimeout(timer);
+		client.close();
+		resolve(null);
+	});
+	req.on("response", headers => {
+		const status = Number(headers[":status"] ?? 0);
+		if (status < 200 || status >= 300) {
+			clearTimeout(timer);
+			client.close();
+			resolve(null);
+		}
+	});
+
+	if (body.length > 0) {
+		req.end(Buffer.from(body));
+	} else {
+		req.end();
+	}
+
+	return promise;
 }
 
 function normalizeCustomModelIds(customModelIds: readonly string[] | undefined): string[] {
@@ -132,15 +177,6 @@ function createCursorReferenceMap(): Map<string, Model<"cursor-agent">> {
 	return references;
 }
 
-function encodeConnectUnaryMessage(payload: Uint8Array): Uint8Array {
-	const framed = new Uint8Array(5 + payload.length);
-	framed[0] = 0;
-	const view = new DataView(framed.buffer, framed.byteOffset, framed.byteLength);
-	view.setUint32(1, payload.length, false);
-	framed.set(payload, 5);
-	return framed;
-}
-
 function decodeGetUsableModelsResponse(payload: Uint8Array) {
 	if (payload.length === 0) {
 		return null;