gsd-pi 2.27.0 → 2.28.0-dev.346ee62

package/packages/pi-ai/src/models.generated.ts

@@ -13388,27 +13388,28 @@ export const MODELS = {
 		"qwen3.5-plus": {
 			id: "qwen3.5-plus",
 			name: "Qwen3.5 Plus",
-			api: "anthropic-messages",
+			api: "openai-completions",
 			provider: "alibaba-coding-plan",
-			baseUrl: "https://coding-intl.dashscope.aliyuncs.com/apps/anthropic",
+			baseUrl: "https://coding-intl.dashscope.aliyuncs.com/v1",
 			reasoning: true,
-			input: ["text", "image"],
+			input: ["text"],
 			cost: {
 				input: 0,
 				output: 0,
 				cacheRead: 0,
 				cacheWrite: 0,
 			},
-			contextWindow: 1000000,
+			contextWindow: 983616,
 			maxTokens: 65536,
-		} satisfies Model<"anthropic-messages">,
+			compat: { thinkingFormat: "qwen", supportsDeveloperRole: false },
+		} satisfies Model<"openai-completions">,
 		"qwen3-max-2026-01-23": {
 			id: "qwen3-max-2026-01-23",
 			name: "Qwen3 Max 2026-01-23",
-			api: "anthropic-messages",
+			api: "openai-completions",
 			provider: "alibaba-coding-plan",
-			baseUrl: "https://coding-intl.dashscope.aliyuncs.com/apps/anthropic",
-			reasoning: false,
+			baseUrl: "https://coding-intl.dashscope.aliyuncs.com/v1",
+			reasoning: true,
 			input: ["text"],
 			cost: {
 				input: 0,
@@ -13416,15 +13417,16 @@ export const MODELS = {
 				cacheRead: 0,
 				cacheWrite: 0,
 			},
-			contextWindow: 262144,
+			contextWindow: 258048,
 			maxTokens: 32768,
-		} satisfies Model<"anthropic-messages">,
+			compat: { thinkingFormat: "qwen", supportsDeveloperRole: false },
+		} satisfies Model<"openai-completions">,
 		"qwen3-coder-next": {
 			id: "qwen3-coder-next",
 			name: "Qwen3 Coder Next",
-			api: "anthropic-messages",
+			api: "openai-completions",
 			provider: "alibaba-coding-plan",
-			baseUrl: "https://coding-intl.dashscope.aliyuncs.com/apps/anthropic",
+			baseUrl: "https://coding-intl.dashscope.aliyuncs.com/v1",
 			reasoning: false,
 			input: ["text"],
 			cost: {
@@ -13433,15 +13435,16 @@ export const MODELS = {
 				cacheRead: 0,
 				cacheWrite: 0,
 			},
-			contextWindow: 262144,
+			contextWindow: 204800,
 			maxTokens: 65536,
-		} satisfies Model<"anthropic-messages">,
+			compat: { supportsDeveloperRole: false },
+		} satisfies Model<"openai-completions">,
 		"qwen3-coder-plus": {
 			id: "qwen3-coder-plus",
 			name: "Qwen3 Coder Plus",
-			api: "anthropic-messages",
+			api: "openai-completions",
 			provider: "alibaba-coding-plan",
-			baseUrl: "https://coding-intl.dashscope.aliyuncs.com/apps/anthropic",
+			baseUrl: "https://coding-intl.dashscope.aliyuncs.com/v1",
 			reasoning: false,
 			input: ["text"],
 			cost: {
@@ -13450,15 +13453,16 @@ export const MODELS = {
 				cacheRead: 0,
 				cacheWrite: 0,
 			},
-			contextWindow: 1000000,
+			contextWindow: 997952,
 			maxTokens: 65536,
-		} satisfies Model<"anthropic-messages">,
+			compat: { supportsDeveloperRole: false },
+		} satisfies Model<"openai-completions">,
 		"MiniMax-M2.5": {
 			id: "MiniMax-M2.5",
 			name: "MiniMax M2.5",
-			api: "anthropic-messages",
+			api: "openai-completions",
 			provider: "alibaba-coding-plan",
-			baseUrl: "https://coding-intl.dashscope.aliyuncs.com/apps/anthropic",
+			baseUrl: "https://coding-intl.dashscope.aliyuncs.com/v1",
 			reasoning: true,
 			input: ["text"],
 			cost: {
@@ -13468,14 +13472,15 @@ export const MODELS = {
 				cacheWrite: 0,
 			},
 			contextWindow: 196608,
-			maxTokens: 24576,
-		} satisfies Model<"anthropic-messages">,
+			maxTokens: 65536,
+			compat: { supportsStore: false, supportsDeveloperRole: false, supportsReasoningEffort: true, maxTokensField: "max_tokens" },
+		} satisfies Model<"openai-completions">,
 		"glm-5": {
 			id: "glm-5",
 			name: "GLM-5",
-			api: "anthropic-messages",
+			api: "openai-completions",
 			provider: "alibaba-coding-plan",
-			baseUrl: "https://coding-intl.dashscope.aliyuncs.com/apps/anthropic",
+			baseUrl: "https://coding-intl.dashscope.aliyuncs.com/v1",
 			reasoning: true,
 			input: ["text"],
 			cost: {
@@ -13486,13 +13491,14 @@ export const MODELS = {
 			},
 			contextWindow: 202752,
 			maxTokens: 16384,
-		} satisfies Model<"anthropic-messages">,
+			compat: { thinkingFormat: "qwen", supportsDeveloperRole: false },
+		} satisfies Model<"openai-completions">,
 		"glm-4.7": {
 			id: "glm-4.7",
 			name: "GLM-4.7",
-			api: "anthropic-messages",
+			api: "openai-completions",
 			provider: "alibaba-coding-plan",
-			baseUrl: "https://coding-intl.dashscope.aliyuncs.com/apps/anthropic",
+			baseUrl: "https://coding-intl.dashscope.aliyuncs.com/v1",
 			reasoning: true,
 			input: ["text"],
 			cost: {
@@ -13501,26 +13507,28 @@ export const MODELS = {
 				cacheRead: 0,
 				cacheWrite: 0,
 			},
-			contextWindow: 202752,
+			contextWindow: 169984,
 			maxTokens: 16384,
-		} satisfies Model<"anthropic-messages">,
+			compat: { thinkingFormat: "qwen", supportsDeveloperRole: false },
+		} satisfies Model<"openai-completions">,
 		"kimi-k2.5": {
 			id: "kimi-k2.5",
 			name: "Kimi K2.5",
-			api: "anthropic-messages",
+			api: "openai-completions",
 			provider: "alibaba-coding-plan",
-			baseUrl: "https://coding-intl.dashscope.aliyuncs.com/apps/anthropic",
+			baseUrl: "https://coding-intl.dashscope.aliyuncs.com/v1",
 			reasoning: true,
-			input: ["text", "image"],
+			input: ["text"],
 			cost: {
 				input: 0,
 				output: 0,
 				cacheRead: 0,
 				cacheWrite: 0,
 			},
-			contextWindow: 262144,
+			contextWindow: 258048,
 			maxTokens: 32768,
-		} satisfies Model<"anthropic-messages">,
+			compat: { thinkingFormat: "zai", supportsDeveloperRole: false },
+		} satisfies Model<"openai-completions">,
 	},
 	"ollama-cloud": {
 		"cogito-2.1:671b": {

package/packages/pi-ai/src/providers/anthropic.ts

@@ -709,10 +709,11 @@ function buildParams(
 	options?: AnthropicOptions,
 ): MessageCreateParamsStreaming {
 	const { cacheControl } = getCacheControl(model.baseUrl, options?.cacheRetention);
-	// For OAuth (Claude Max/Pro), strip variant suffixes like [1m] from model ID.
+	// Strip variant suffixes like [1m] from model ID before sending to the API.
 	// The API only accepts the base model ID (e.g. "claude-opus-4-6"),
 	// not internal variant identifiers (e.g. "claude-opus-4-6[1m]").
-	const apiModelId = isOAuthToken ? model.id.replace(/\[.*\]$/, "") : model.id;
+	// This applies to all auth methods — API keys, OAuth, and Copilot alike.
+	const apiModelId = model.id.replace(/\[.*\]$/, "");
 	const params: MessageCreateParamsStreaming = {
 		model: apiModelId,
 		messages: convertMessages(context.messages, model, isOAuthToken, cacheControl),

package/packages/pi-ai/src/providers/openai-codex-responses.ts

@@ -368,9 +368,14 @@ async function* mapCodexEvents(events: AsyncIterable<Record<string, unknown>>):
 		if (!type) continue;
 
 		if (type === "error") {
-			const code = (event as { code?: string }).code || "";
-			const message = (event as { message?: string }).message || "";
-			throw new Error(`Codex error: ${message || code || JSON.stringify(event)}`);
+			// Codex error events nest details under event.error (e.g.
+			// { type: "error", error: { type: "server_error", code: "server_error", message: "..." } })
+			const errorObj = (event as { error?: { code?: string; type?: string; message?: string } }).error;
+			const code = errorObj?.code || (event as { code?: string }).code || "";
+			const errorType = errorObj?.type || "";
+			const message = errorObj?.message || (event as { message?: string }).message || "";
+			const prefix = errorType ? `Codex ${errorType}` : "Codex error";
+			throw new Error(`${prefix}: ${message || code || JSON.stringify(event)}`);
 		}
 
 		if (type === "response.failed") {
@@ -410,8 +415,14 @@ async function* parseSSE(response: Response): AsyncGenerator<Record<string, unkn
 	while (true) {
 		const { done, value } = await reader.read();
 		if (done) break;
-		buffer += decoder.decode(value, { stream: true });
 
+		const decoded = decoder.decode(value, { stream: true });
+		// Avoid appending to an empty buffer — assign directly to skip the
+		// string concatenation and its intermediate allocation.
+		buffer = buffer ? buffer + decoded : decoded;
+
+		// Consume all complete SSE messages (delimited by \n\n) so the
+		// buffer only ever holds one partial message between reads.
 		let idx = buffer.indexOf("\n\n");
 		while (idx !== -1) {
 			const chunk = buffer.slice(0, idx);

package/packages/pi-ai/src/utils/oauth/anthropic.ts

@@ -61,6 +61,7 @@ export async function loginAnthropic(
 			redirect_uri: REDIRECT_URI,
 			code_verifier: verifier,
 		}),
+		signal: AbortSignal.timeout(30_000),
 	});
 
 	if (!tokenResponse.ok) {
@@ -97,6 +98,7 @@ export async function refreshAnthropicToken(refreshToken: string): Promise<OAuth
 			client_id: CLIENT_ID,
 			refresh_token: refreshToken,
 		}),
+		signal: AbortSignal.timeout(30_000),
 	});
 
 	if (!response.ok) {

package/packages/pi-ai/src/utils/oauth/github-copilot.ts

@@ -89,7 +89,10 @@ export function getGitHubCopilotBaseUrl(token?: string, enterpriseDomain?: strin
 }
 
 async function fetchJson(url: string, init: RequestInit): Promise<unknown> {
-	const response = await fetch(url, init);
+	const response = await fetch(url, {
+		...init,
+		signal: init.signal ?? AbortSignal.timeout(30_000),
+	});
 	if (!response.ok) {
 		const text = await response.text();
 		throw new Error(`${response.status} ${response.statusText}: ${text}`);
@@ -276,6 +279,7 @@ async function enableGitHubCopilotModel(token: string, modelId: string, enterpri
 				"x-interaction-type": "chat-policy",
 			},
 			body: JSON.stringify({ state: "enabled" }),
+			signal: AbortSignal.timeout(30_000),
 		});
 		return response.ok;
 	} catch {

package/packages/pi-ai/src/utils/oauth/google-antigravity.ts

@@ -184,6 +184,7 @@ async function discoverProject(accessToken: string, onProgress?: (message: strin
 						pluginType: "GEMINI",
 					},
 				}),
+				signal: AbortSignal.timeout(30_000),
 			});
 
 			if (loadResponse.ok) {
@@ -220,6 +221,7 @@ async function getUserEmail(accessToken: string): Promise<string | undefined> {
 			headers: {
 				Authorization: `Bearer ${accessToken}`,
 			},
+			signal: AbortSignal.timeout(30_000),
 		});
 
 		if (response.ok) {
@@ -245,6 +247,7 @@ export async function refreshAntigravityToken(refreshToken: string, projectId: s
 			refresh_token: refreshToken,
 			grant_type: "refresh_token",
 		}),
+		signal: AbortSignal.timeout(30_000),
 	});
 
 	if (!response.ok) {
@@ -392,6 +395,7 @@ export async function loginAntigravity(
 				redirect_uri: REDIRECT_URI,
 				code_verifier: verifier,
 			}),
+			signal: AbortSignal.timeout(30_000),
 		});
 
 		if (!tokenResponse.ok) {

package/packages/pi-ai/src/utils/oauth/google-gemini-cli.ts

@@ -207,6 +207,7 @@ async function pollOperation(
 		const response = await fetch(`${CODE_ASSIST_ENDPOINT}/v1internal/${operationName}`, {
 			method: "GET",
 			headers,
+			signal: AbortSignal.timeout(30_000),
 		});
 
 		if (!response.ok) {
@@ -250,6 +251,7 @@ async function discoverProject(accessToken: string, onProgress?: (message: strin
 				duetProject: envProjectId,
 			},
 		}),
+		signal: AbortSignal.timeout(30_000),
 	});
 
 	let data: LoadCodeAssistPayload;
@@ -321,6 +323,7 @@ async function discoverProject(accessToken: string, onProgress?: (message: strin
 		method: "POST",
 		headers,
 		body: JSON.stringify(onboardBody),
+		signal: AbortSignal.timeout(30_000),
 	});
 
 	if (!onboardResponse.ok) {
@@ -362,6 +365,7 @@ async function getUserEmail(accessToken: string): Promise<string | undefined> {
 			headers: {
 				Authorization: `Bearer ${accessToken}`,
 			},
+			signal: AbortSignal.timeout(30_000),
 		});
 
 		if (response.ok) {
@@ -387,6 +391,7 @@ export async function refreshGoogleCloudToken(refreshToken: string, projectId: s
 			refresh_token: refreshToken,
 			grant_type: "refresh_token",
 		}),
+		signal: AbortSignal.timeout(30_000),
 	});
 
 	if (!response.ok) {
@@ -534,6 +539,7 @@ export async function loginGeminiCli(
 				redirect_uri: REDIRECT_URI,
 				code_verifier: verifier,
 			}),
+			signal: AbortSignal.timeout(30_000),
 		});
 
 		if (!tokenResponse.ok) {

package/packages/pi-ai/src/utils/oauth/index.ts

@@ -126,8 +126,8 @@ export async function getOAuthApiKey(
 	if (Date.now() >= creds.expires) {
 		try {
 			creds = await provider.refreshToken(creds);
-		} catch (_error) {
-			throw new Error(`Failed to refresh OAuth token for ${providerId}`);
+		} catch (error) {
+			throw new Error(`Failed to refresh OAuth token for ${providerId}`, { cause: error });
 		}
 	}
 

package/packages/pi-ai/src/utils/oauth/openai-codex.ts

@@ -114,6 +114,7 @@ async function exchangeAuthorizationCode(
 			code_verifier: verifier,
 			redirect_uri: redirectUri,
 		}),
+		signal: AbortSignal.timeout(30_000),
 	});
 
 	if (!response.ok) {
@@ -151,6 +152,7 @@ async function refreshAccessToken(refreshToken: string): Promise<TokenResult> {
 				refresh_token: refreshToken,
 				client_id: CLIENT_ID,
 			}),
+			signal: AbortSignal.timeout(30_000),
 		});
 
 		if (!response.ok) {

package/packages/pi-coding-agent/dist/core/agent-session.js

@@ -1936,7 +1936,7 @@ export class AgentSession {
             return false;
         const err = message.errorMessage;
         // Match: overloaded_error, rate limit, 429, 500, 502, 503, 504, service unavailable, connection errors, fetch failed, terminated, retry delay exceeded, network unavailable / auth expired (transient network failures)
-        return /overloaded|rate.?limit|too many requests|429|500|502|503|504|service.?unavailable|server error|internal error|connection.?error|connection.?refused|other side closed|fetch failed|upstream.?connect|reset before headers|terminated|retry delay|network.?(?:is\s+)?unavailable|credentials.*expired|temporarily backed off/i.test(err);
+        return /overloaded|rate.?limit|too many requests|429|500|502|503|504|service.?unavailable|server.?error|internal.?error|connection.?error|connection.?refused|other side closed|fetch failed|upstream.?connect|reset before headers|terminated|retry delay|network.?(?:is\s+)?unavailable|credentials.*expired|temporarily backed off/i.test(err);
     }
     /**
      * Classify an error message into a usage-limit error type for credential backoff.