@gajae-code/ai 0.15.6 → 0.16.1

package/src/provider-models/openai-compat.ts

@@ -59,15 +59,39 @@ function toInputCapabilities(value: unknown): ("text" | "image")[] {
 	return supportsImage ? ["text", "image"] : ["text"];
 }
 
-async function fetchModelsDevPayload(fetchImpl: typeof fetch = fetch): Promise<unknown> {
-	const response = await fetchImpl(MODELS_DEV_URL, {
-		method: "GET",
-		headers: { Accept: "application/json" },
-	});
-	if (!response.ok) {
-		throw new Error(`models.dev fetch failed: ${response.status}`);
+/**
+ * models.dev serves one catalog document describing every provider, and a
+ * discovery pass resolves several providers from it. Downloads are coalesced
+ * per fetch implementation so one pass transfers the catalog once instead of
+ * once per provider; the window is short enough that a later refresh still
+ * observes catalog updates.
+ */
+const MODELS_DEV_PAYLOAD_TTL_MS = 60_000;
+const modelsDevPayloadCache = new WeakMap<typeof fetch, { at: number; payload: Promise<unknown> }>();
+
+export async function fetchModelsDevPayload(fetchImpl: typeof fetch = fetch): Promise<unknown> {
+	const now = Date.now();
+	const cached = modelsDevPayloadCache.get(fetchImpl);
+	if (cached && now - cached.at < MODELS_DEV_PAYLOAD_TTL_MS) return cached.payload;
+	const payload = (async () => {
+		const response = await fetchImpl(MODELS_DEV_URL, {
+			method: "GET",
+			headers: { Accept: "application/json" },
+			signal: AbortSignal.timeout(5_000),
+		});
+		if (!response.ok) {
+			throw new Error(`models.dev fetch failed: ${response.status}`);
+		}
+		return (await response.json()) as unknown;
+	})();
+	const entry = { at: now, payload };
+	modelsDevPayloadCache.set(fetchImpl, entry);
+	try {
+		return await payload;
+	} catch (error) {
+		if (modelsDevPayloadCache.get(fetchImpl) === entry) modelsDevPayloadCache.delete(fetchImpl);
+		throw error;
 	}
-	return response.json();
 }
 
 function anthropicToolChoiceCompat(modelId: string): Pick<Model<"anthropic-messages">, "compat"> {
@@ -2357,11 +2381,15 @@ const OPENCODE_GO_BASE_PATH = "https://opencode.ai/zen/go";
 const OPENCODE_ZEN_API_RESOLUTION = createOpenCodeApiResolution("https://opencode.ai/zen");
 const OPENCODE_GO_CHAT_COMPLETIONS_MODEL_IDS = [
 	"deepseek-v4-flash",
+	"deepseek-v4-flash-vision-exp",
 	"deepseek-v4-pro",
 	"glm-5.1",
 	"glm-5.2",
+	"glm-5.3-flash",
 	"kimi-k2.6",
 	"kimi-k2.7-code",
+	"hy4-preview",
+	"longcat-2.0",
 	"mimo-v2.5",
 	"mimo-v2.5-pro",
 ] as const;
@@ -2372,6 +2400,7 @@ const OPENCODE_GO_MESSAGES_MODEL_IDS = [
 	"qwen3.6-plus",
 	"qwen3.7-max",
 	"qwen3.7-plus",
+	"qwen3.8-flash",
 ] as const;
 const OPENCODE_GO_API_OVERRIDES: Readonly<Record<string, Api>> = {
 	...Object.fromEntries(OPENCODE_GO_CHAT_COMPLETIONS_MODEL_IDS.map(id => [id, "openai-completions"])),
@@ -2796,6 +2825,62 @@ const OPENCODE_GO_OFFICIAL_MODELS: Readonly<Record<string, OpenCodeGoOfficialMod
 		reasoning: true,
 		cost: { input: 0.066, output: 0.26, cacheRead: 0.029, cacheWrite: 0 },
 	},
+	"deepseek-v4-flash-vision-exp": {
+		name: "DeepSeek V4 Flash Vision Exp",
+		contextWindow: 1_000_000,
+		maxTokens: 384_000,
+		input: ["text", "image"],
+		reasoning: true,
+		cost: { input: 0.22, output: 0.66, cacheRead: 0.007, cacheWrite: 0 },
+	},
+	"glm-5.3-flash": {
+		name: "GLM-5.3-Flash (2x usage)",
+		contextWindow: 1_000_000,
+		maxTokens: 131_072,
+		input: ["text", "image"],
+		reasoning: true,
+		cost: { input: 0.075, output: 0.25, cacheRead: 0.015, cacheWrite: 0 },
+	},
+	"grok-4.6": {
+		name: "Grok 4.6",
+		contextWindow: 500_000,
+		maxTokens: 500_000,
+		input: ["text", "image"],
+		reasoning: true,
+		cost: { input: 2, output: 6, cacheRead: 0.5, cacheWrite: 0 },
+	},
+	"hy4-preview": {
+		name: "Hy4 preview",
+		contextWindow: 1_024_000,
+		maxTokens: 64_000,
+		input: ["text"],
+		reasoning: true,
+		cost: { input: 0.834, output: 2.501, cacheRead: 0.042, cacheWrite: 0 },
+	},
+	"longcat-2.0": {
+		name: "LongCat-2.0",
+		contextWindow: 1_000_000,
+		maxTokens: 131_072,
+		input: ["text"],
+		reasoning: true,
+		cost: { input: 0.3, output: 1.2, cacheRead: 0.006, cacheWrite: 0 },
+	},
+	"muse-spark-1.2-contributor": {
+		name: "Muse Spark 1.2 Contributor",
+		contextWindow: 1_048_576,
+		maxTokens: 131_072,
+		input: ["text", "image"],
+		reasoning: true,
+		cost: { input: 0.1, output: 0.2, cacheRead: 0.002, cacheWrite: 0 },
+	},
+	"qwen3.8-flash": {
+		name: "Qwen3.8 Flash",
+		contextWindow: 1_000_000,
+		maxTokens: 131_072,
+		input: ["text", "image"],
+		reasoning: true,
+		cost: { input: 0.15, output: 0.47, cacheRead: 0.016, cacheWrite: 0.2 },
+	},
 };
 
 function applyOpenCodeGoOfficialMetadata<TApi extends Api>(model: Model<TApi>): Model<TApi> {

package/src/providers/amazon-bedrock.ts

@@ -8,6 +8,7 @@
  */
 
 import { $credentialEnv, $env, $flag, extractHttpStatusFromError, fetchWithRetry } from "@gajae-code/utils";
+import { assertAwsRegionLabel } from "../adapter-internals/aws-region";
 import type { Effort } from "../model-thinking";
 import {
 	mapEffortToAnthropicAdaptiveEffort,
@@ -202,9 +203,10 @@ export const streamBedrock: StreamFunction<"bedrock-converse-stream"> = (
 
 		const blocks = output.content as Block[];
 		let rawRequestDump: RawHttpRequestDump | undefined;
-		const region = options.region || $env.AWS_REGION || $env.AWS_DEFAULT_REGION || "us-east-1";
+		const region = options.region ?? $env.AWS_REGION ?? $env.AWS_DEFAULT_REGION ?? "us-east-1";
 
 		try {
+			assertAwsRegionLabel(region);
 			const cacheRetention = resolveCacheRetention(options.cacheRetention);
 			const resolvedToolChoice = resolveToolChoice(model, options.toolChoice);
 			const toolConfig = convertToolConfig(context.tools, resolvedToolChoice.resolvedChoice);
@@ -303,6 +305,7 @@ export const streamBedrock: StreamFunction<"bedrock-converse-stream"> = (
 					method: "POST",
 					headers: await buildRequestHeaders(retryBody),
 					body: retryBody,
+					redirect: "error",
 					signal: options.signal,
 					maxAttempts: 1,
 				});
@@ -312,6 +315,7 @@ export const streamBedrock: StreamFunction<"bedrock-converse-stream"> = (
 				method: "POST",
 				headers: requestHeaders,
 				body,
+				redirect: "error",
 				signal: options.signal,
 				maxAttempts: resolveRetryBudget(options.requestMaxRetries, 4) + 1,
 			});

package/src/providers/anthropic.d.ts

@@ -107,7 +107,7 @@ export interface CpaToolAliasRestoreFailure {
  */
 export declare function parseCpaToolAliasRestoreFailure(error: unknown): CpaToolAliasRestoreFailure | undefined;
 export declare function isCpaToolAliasRestoreFailure(error: unknown): boolean;
-export declare const claudeCodeVersion = "2.1.219";
+export declare const claudeCodeVersion = "2.1.257";
 export declare const claudeCodeEntrypoint = "sdk-cli";
 export declare const claudeToolPrefix: string;
 export declare const claudeCodeSystemInstruction = "You are a Claude agent, built on Anthropic's Claude Agent SDK.";

package/src/providers/anthropic.ts

@@ -828,7 +828,7 @@ function getCacheControl(
 }
 
 // Stealth mode: Mimic Anthropic Code headers and tool prefixing.
-export const claudeCodeVersion = "2.1.219";
+export const claudeCodeVersion = "2.1.257";
 export const claudeCodeEntrypoint = "sdk-cli";
 export const claudeToolPrefix: string = "proxy_";
 export const claudeCodeSystemInstruction = "You are a Claude agent, built on Anthropic's Claude Agent SDK.";

package/src/providers/aws-credentials.ts

@@ -22,6 +22,7 @@
 import * as fs from "node:fs";
 import * as path from "node:path";
 import { $env, getTrustedHomeDir, isEnoent, logger } from "@gajae-code/utils";
+import { assertAwsRegionLabel } from "../adapter-internals/aws-region";
 import {
 	type AwsIniFile,
 	classifyAwsProfileCapability,
@@ -155,6 +156,7 @@ async function readSsoCredentials(
 		}
 	}
 	if (!startUrl || !ssoRegion) return undefined;
+	assertAwsRegionLabel(ssoRegion);
 
 	const token = await loadSsoCachedToken(startUrl, sessionName);
 	if (!token?.accessToken) {
@@ -172,6 +174,7 @@ async function readSsoCredentials(
 	const response = await fetch(url, {
 		method: "GET",
 		headers: { "x-amz-sso_bearer_token": token.accessToken },
+		redirect: "error",
 		signal,
 	});
 	if (!response.ok) {
@@ -411,6 +414,7 @@ async function readImdsCredentials(parentSignal: AbortSignal | undefined): Promi
 		const tokenRes = await fetch(`http://${IMDS_HOST}/latest/api/token`, {
 			method: "PUT",
 			headers: { "x-aws-ec2-metadata-token-ttl-seconds": "21600" },
+			redirect: "error",
 			signal,
 		});
 		if (!tokenRes.ok) return undefined;
@@ -418,6 +422,7 @@ async function readImdsCredentials(parentSignal: AbortSignal | undefined): Promi
 
 		const roleRes = await fetch(`http://${IMDS_HOST}/latest/meta-data/iam/security-credentials/`, {
 			headers: { "x-aws-ec2-metadata-token": token },
+			redirect: "error",
 			signal,
 		});
 		if (!roleRes.ok) return undefined;
@@ -428,6 +433,7 @@ async function readImdsCredentials(parentSignal: AbortSignal | undefined): Promi
 			`http://${IMDS_HOST}/latest/meta-data/iam/security-credentials/${encodeURIComponent(role)}`,
 			{
 				headers: { "x-aws-ec2-metadata-token": token },
+				redirect: "error",
 				signal,
 			},
 		);

package/src/providers/cursor.d.ts

@@ -1,8 +1,9 @@
 import { type JsonValue } from "@bufbuild/protobuf";
-import type { CursorExecHandlerResult, CursorExecHandlers, CursorToolResultHandler, Message, Model, StreamFunction, StreamOptions, ToolCall, ToolResultMessage } from "../types";
+import type { AssistantMessage, CursorExecHandlerResult, CursorExecHandlers, CursorToolResultHandler, Message, Model, StreamFunction, StreamOptions, ToolCall, ToolResultMessage, Usage } from "../types";
 import { kCursorExecResolved } from "../utils/block-symbols";
 import { CURSOR_CLIENT_VERSION } from "./cursor/client-version";
 import type { CursorRule, RequestedModel_ModelParameterbytes } from "./cursor/gen/agent_pb";
+import { type ConversationStateStructure } from "./cursor/gen/agent_pb";
 export declare const CURSOR_API_URL = "https://api2.cursor.sh";
 export { CURSOR_CLIENT_VERSION };
 /** Drop all cached state + blob bytes for a conversation (F15 bound + session-teardown hook). */
@@ -31,6 +32,19 @@ type ToolCallState = ToolCall & {
     kind: "mcp" | "todo_write" | "native" | "cursor-exec";
     [kCursorExecResolved]?: true;
 };
+interface UsageState {
+    sawTokenDelta: boolean;
+    /**
+     * Latest `ConversationTokenDetails.used_tokens`: the whole conversation's
+     * token consumption as counted by Cursor, not this turn's output.
+     */
+    conversationUsedTokens: number;
+    /** Output tokens already included in the latest checkpoint snapshot. */
+    checkpointOutputTokens: number;
+    /** Whether the current stream received a checkpoint, including an explicit zero. */
+    hasConversationCheckpoint: boolean;
+    pendingCheckpoint?: ConversationStateStructure;
+}
 /** Exported for tests: verifies handler is invoked with correct `this` when passed as bound. */
 export declare function resolveExecHandler<TArgs, TResult>(args: TArgs, handler: ((args: TArgs) => Promise<CursorExecHandlerResult<TResult>>) | undefined, onToolResult: CursorToolResultHandler | undefined, buildFromToolResult: (toolResult: ToolResultMessage) => TResult, buildRejected: (reason: string) => TResult, buildError: (error: string) => TResult): Promise<{
     execResult: TResult;
@@ -44,6 +58,18 @@ export declare function createCursorMessageQueueForTest(onError?: (error: unknow
 /** Exported for direct regression coverage of the JSON-safety boundary. */
 export declare function cursorJsonSafeValueForTest(value: unknown): unknown;
 export declare function buildNativeToolCallBlock(toolCall: Record<string, unknown>, callId: string, index: number): ToolCallState | null;
+/**
+ * Cursor streams output tokens as deltas and reports whole-conversation
+ * consumption separately as `ConversationTokenDetails.used_tokens`. Derive
+ * prompt tokens from the difference so context accounting and compaction see a
+ * real prompt size instead of zero.
+ */
+export declare function finalizeCursorUsage(output: AssistantMessage, usageState: UsageState): void;
+/** Exposes {@link finalizeCursorUsage} for tests without a live HTTP/2 stream. */
+export declare function finalizeCursorUsageForTest(usedTokens: number, outputTokens: number, options?: {
+    checkpointOutputTokens?: number;
+    hasConversationCheckpoint?: boolean;
+}): Usage;
 /**
  * Build `ConversationStateStructure.rootPromptMessagesJson` blob IDs for the
  * system prompt plus prior conversation history, as JSON blobs matching