kiro-proxy 0.1.15

package/README.md

@@ -0,0 +1,90 @@
+[English](README_EN.md) | 中文
+
+# kiro-proxy
+
+让 [Kiro](https://kiro.dev) 订阅内含的 Claude 模型可以在 Claude Code 中使用。
+
+通过读取 Kiro 的认证 token，代理请求到 Amazon Q Developer，暴露 OpenAI 和 Anthropic 兼容的 API 接口。
+
+## 前提
+
+需要先安装并登录 Kiro，确保 `~/.aws/sso/cache/kiro-auth-token.json` 存在且未过期。
+
+## 快速开始
+
+```bash
+npx @colin3191/kiro-proxy
+```
+
+服务默认监听 `http://localhost:3456`。
+
+## 配置
+
+| 环境变量 | 默认值 | 说明 |
+|----------|--------|------|
+| `PORT`   | `3456` | 监听端口 |
+
+## API
+
+### GET /v1/models — 查询可用模型
+
+```bash
+curl http://localhost:3456/v1/models
+```
+
+### POST /v1/chat/completions — OpenAI 兼容
+
+```bash
+# 非流式
+curl http://localhost:3456/v1/chat/completions \
+  -H "Content-Type: application/json" \
+  -d '{"model": "claude-sonnet-4.6", "messages": [{"role": "user", "content": "Hello"}]}'
+
+# 流式
+curl http://localhost:3456/v1/chat/completions \
+  -H "Content-Type: application/json" \
+  -d '{"model": "claude-sonnet-4.6", "messages": [{"role": "user", "content": "Hello"}], "stream": true}'
+```
+
+### POST /v1/messages — Anthropic 兼容
+
+```bash
+# 非流式
+curl http://localhost:3456/v1/messages \
+  -H "Content-Type: application/json" \
+  -H "x-api-key: any" \
+  -d '{"model": "claude-sonnet-4.6", "max_tokens": 1024, "messages": [{"role": "user", "content": "Hello"}]}'
+
+# 流式
+curl http://localhost:3456/v1/messages \
+  -H "Content-Type: application/json" \
+  -H "x-api-key: any" \
+  -d '{"model": "claude-sonnet-4.6", "max_tokens": 1024, "messages": [{"role": "user", "content": "Hello"}], "stream": true}'
+```
+
+### GET /health
+
+检查 token 状态及过期时间。
+
+## 与 Claude Code 集成
+
+Claude Code 默认使用 Anthropic 官方 model ID，需要通过环境变量映射到 Q Developer 的 model ID。
+
+在 `~/.claude/settings.json` 中添加：
+
+```json
+{
+  "env": {
+    "ANTHROPIC_AUTH_TOKEN": "any",
+    "ANTHROPIC_BASE_URL": "http://localhost:3456",
+    "ANTHROPIC_MODEL": "claude-sonnet-4.6",
+    "ANTHROPIC_DEFAULT_SONNET_MODEL": "claude-sonnet-4.6",
+    "ANTHROPIC_DEFAULT_OPUS_MODEL": "claude-opus-4.6",
+    "ANTHROPIC_DEFAULT_HAIKU_MODEL": "claude-haiku-4.5"
+  }
+}
+```
+
+## 相关项目
+
+- [kiro-web-search](https://github.com/Colin3191/kiro-web-search) — 将 Kiro 内置的联网搜索封装为 MCP server，可在 Claude Code 等客户端中使用

package/logger.js

@@ -0,0 +1,82 @@
+import crypto from 'crypto';
+
+export const c = {
+  reset: '\x1b[0m',
+  dim: '\x1b[2m',
+  cyan: '\x1b[36m',
+  green: '\x1b[32m',
+  yellow: '\x1b[33m',
+  blue: '\x1b[34m',
+  magenta: '\x1b[35m',
+  gray: '\x1b[90m',
+  red: '\x1b[31m',
+};
+
+const METHOD_COLORS = {
+  POST: c.green,
+  GET: c.blue,
+  DELETE: c.yellow,
+};
+
+function timestamp() {
+  const now = new Date();
+  return c.gray + now.toLocaleTimeString('en-GB', { hour12: false }) + '.' + String(now.getMilliseconds()).padStart(3, '0') + c.reset;
+}
+
+/**
+ * 生成 3 字符的短请求 ID
+ */
+export function reqId() {
+  return crypto.randomBytes(2).toString('hex').slice(0, 3);
+}
+
+/**
+ * HTTP 请求日志（server.js 用）
+ * 格式: 12:34:56.789 POST /v1/messages [a3f] model=xxx
+ */
+export function log(method, path, id, info) {
+  const methodColor = METHOD_COLORS[method] || c.magenta;
+  const parts = [timestamp()];
+  if (id) parts.push(c.magenta + `[${id}]` + c.reset);
+  parts.push(methodColor + method + c.reset, c.cyan + path + c.reset);
+  if (info) parts.push(c.dim + (typeof info === 'string' ? info : JSON.stringify(info)) + c.reset);
+  console.log(parts.join(' '));
+}
+
+/**
+ * 带标签的日志（q-client.js / token-reader.js 用）
+ * 格式: 12:34:56.789 [token] Refreshing Social token...
+ */
+export function tagLog(tag, ...args) {
+  const msg = args.map(a => typeof a === 'string' ? a : JSON.stringify(a)).join(' ');
+  console.log(`${timestamp()} ${c.cyan}[${tag}]${c.reset} ${msg}`);
+}
+
+export function tagWarn(tag, ...args) {
+  const msg = args.map(a => typeof a === 'string' ? a : JSON.stringify(a)).join(' ');
+  console.warn(`${timestamp()} ${c.yellow}[${tag}]${c.reset} ${msg}`);
+}
+
+export function tagError(tag, ...args) {
+  const msg = args.map(a => typeof a === 'string' ? a : JSON.stringify(a)).join(' ');
+  console.error(`${timestamp()} ${c.red}[${tag}]${c.reset} ${msg}`);
+}
+
+/**
+ * 请求完成汇总行
+ * 格式: 12:34:56.789  └─ [a3f] 4.07s context=0.21% 0.0096 credits
+ */
+export function logSummary(id, elapsed, stats) {
+  const duration = elapsed >= 1000 ? `${(elapsed / 1000).toFixed(2)}s` : `${elapsed}ms`;
+  const parts = [];
+  parts.push(duration);
+  if (stats.context) parts.push(`context=${stats.context}`);
+  if (stats.metering) parts.push(stats.metering);
+  if (stats.tokens) parts.push(stats.tokens);
+  if (stats.estTokens) parts.push(stats.estTokens);
+  if (stats.links) parts.push(stats.links);
+  if (stats.invalid) parts.push(`invalid: ${stats.invalid}`);
+  if (stats.codeRef) parts.push(`codeRef: ${typeof stats.codeRef === 'string' ? stats.codeRef : JSON.stringify(stats.codeRef)}`);
+  const tag = id ? `${c.magenta}[${id}]${c.reset} ` : '';
+  console.log(`${timestamp()} ${tag}${c.dim}${parts.join(' | ')}${c.reset}`);
+}

package/package.json

@@ -0,0 +1,28 @@
+{
+  "name": "kiro-proxy",
+  "version": "0.1.15",
+  "description": "Kiro API proxy with OpenAI and Anthropic compatible endpoints",
+  "type": "module",
+  "bin": {
+    "kiro-proxy": "./server.js"
+  },
+  "engines": {
+    "node": ">=18"
+  },
+  "files": [
+    "server.js",
+    "q-client.js",
+    "token-reader.js",
+    "token-counter.js",
+    "usage-tracker.js",
+    "logger.js"
+  ],
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/Colin3191/kiro-proxy.git"
+  },
+  "dependencies": {
+    "@aws/codewhisperer-streaming-client": "^1.0.34",
+    "express": "^4.21.0"
+  }
+}

package/q-client.js

@@ -0,0 +1,501 @@
+import { CodeWhispererStreaming, GenerateAssistantResponseCommand } from '@aws/codewhisperer-streaming-client';
+import crypto from 'crypto';
+import os from 'os';
+
+// region → endpoint 映射
+const REGION_ENDPOINTS = {
+  'us-east-1': 'https://q.us-east-1.amazonaws.com',
+  'eu-west-1': 'https://q.eu-west-1.amazonaws.com',
+  'ap-southeast-1': 'https://q.ap-southeast-1.amazonaws.com',
+  'ap-northeast-1': 'https://q.ap-northeast-1.amazonaws.com',
+  'eu-central-1': 'https://q.eu-central-1.amazonaws.com',
+  'ap-south-1': 'https://q.ap-south-1.amazonaws.com',
+  'ca-central-1': 'https://q.ca-central-1.amazonaws.com',
+};
+const DEFAULT_REGION = 'us-east-1';
+const KIRO_VERSION = process.env.KIRO_VERSION || '0.11.107';
+
+function buildUserAgent(machineId) {
+  return `KiroIDE ${KIRO_VERSION} ${machineId || os.hostname()}`;
+}
+
+function regionFromArn(arn) {
+  if (!arn) return null;
+  const parts = arn.split(':');
+  return parts.length >= 4 ? parts[3] : null;
+}
+
+function endpointForRegion(region) {
+  return REGION_ENDPOINTS[region] || `https://q.${region}.amazonaws.com`;
+}
+
+function addRequiredHeaders(client, { agentMode = 'vibe', optOut = true, authMethod, provider } = {}) {
+  if (optOut) {
+    client.middlewareStack.add(
+      (next) => async (args) => {
+        args.request.headers = { ...args.request.headers, 'x-amzn-codewhisperer-optout': 'true' };
+        return next(args);
+      },
+      { step: 'build', name: 'optOutHeader' }
+    );
+  }
+  client.middlewareStack.add(
+    (next) => async (args) => {
+      args.request.headers = { ...args.request.headers, 'x-amzn-kiro-agent-mode': agentMode };
+      return next(args);
+    },
+    { step: 'build', name: 'agentModeHeader' }
+  );
+  if (authMethod === 'external_idp') {
+    client.middlewareStack.add(
+      (next) => async (args) => {
+        args.request.headers = { ...args.request.headers, TokenType: 'EXTERNAL_IDP' };
+        return next(args);
+      },
+      { step: 'build', name: 'tokenTypeHeader' }
+    );
+  }
+  if (provider === 'Internal') {
+    client.middlewareStack.add(
+      (next) => async (args) => {
+        args.request.headers = { ...args.request.headers, 'redirect-for-internal': 'true' };
+        return next(args);
+      },
+      { step: 'build', name: 'redirectForInternal' }
+    );
+  }
+}
+
+export function createClient(accessToken, { endpoint, region, authMethod, profileArn, provider, machineId } = {}) {
+  const arnRegion = regionFromArn(profileArn);
+  const finalRegion = region || arnRegion || DEFAULT_REGION;
+  const finalEndpoint = endpoint || endpointForRegion(finalRegion);
+
+  const client = new CodeWhispererStreaming({
+    region: finalRegion,
+    endpoint: finalEndpoint,
+    token: { token: accessToken },
+    customUserAgent: buildUserAgent(machineId),
+  });
+  addRequiredHeaders(client, { authMethod, provider });
+  return client;
+}
+
+// ============================================================
+// Anthropic tools → CodeWhisperer toolSpecification
+// ============================================================
+function convertTools(tools) {
+  if (!tools || tools.length === 0) return undefined;
+  return tools
+    .filter(t => t.name !== 'web_search' && t.name !== 'websearch')
+    .map(t => ({
+      toolSpecification: {
+        name: t.name,
+        description: (t.description || '').slice(0, 10000),
+        inputSchema: { json: t.input_schema || t.parameters || {} },
+      },
+    }));
+}
+
+// ============================================================
+// Anthropic messages → CodeWhisperer conversationState
+// ============================================================
+
+/**
+ * 从 Anthropic content blocks 中提取图片，转换为 CodeWhisperer 格式
+ * Anthropic 格式: { type: "image", source: { type: "base64", media_type: "image/png", data: "..." } }
+ * CodeWhisperer 格式: { format: "png", source: { bytes: Buffer } }
+ */
+function extractImages(content) {
+  if (!Array.isArray(content)) return [];
+  const formatMap = { 'image/png': 'png', 'image/jpeg': 'jpeg', 'image/gif': 'gif', 'image/webp': 'webp' };
+  const images = [];
+
+  for (const block of content) {
+    if (block.type === 'image' && block.source) {
+      if (block.source.type === 'base64' && block.source.data) {
+        const format = formatMap[block.source.media_type] || 'jpeg';
+        images.push({ format, source: { bytes: Buffer.from(block.source.data, 'base64') } });
+      } else if (block.source.type === 'url' && block.source.url) {
+        // data URL: data:image/png;base64,iVBOR...
+        const url = block.source.url;
+        if (url.startsWith('data:')) {
+          const parts = url.split(',');
+          if (parts.length >= 2) {
+            const mimeMatch = parts[0].match(/data:(image\/\w+)/);
+            const format = mimeMatch ? (formatMap[mimeMatch[1]] || 'jpeg') : 'jpeg';
+            images.push({ format, source: { bytes: Buffer.from(parts[1], 'base64') } });
+          }
+        }
+      }
+    }
+    // LangChain/OpenAI 格式: { type: "image_url", image_url: { url: "data:..." } }
+    if (block.type === 'image_url' && block.image_url) {
+      const url = typeof block.image_url === 'string' ? block.image_url : block.image_url.url;
+      if (url?.startsWith('data:')) {
+        const parts = url.split(',');
+        if (parts.length >= 2) {
+          const mimeMatch = parts[0].match(/data:(image\/\w+)/);
+          const format = mimeMatch ? (formatMap[mimeMatch[1]] || 'jpeg') : 'jpeg';
+          images.push({ format, source: { bytes: Buffer.from(parts[1], 'base64') } });
+        }
+      }
+    }
+  }
+  return images;
+}
+
+/**
+ * 从 Anthropic content blocks 中提取文本
+ */
+function extractText(content) {
+  if (typeof content === 'string') return content;
+  if (!Array.isArray(content)) return '';
+  return content
+    .filter(b => b.type === 'text')
+    .map(b => b.text)
+    .join('');
+}
+
+/**
+ * 从 assistant content blocks 中提取 tool_use 调用
+ */
+function extractToolUses(content) {
+  if (!Array.isArray(content)) return [];
+  return content
+    .filter(b => b.type === 'tool_use')
+    .map(b => ({ toolUseId: b.id, name: b.name, input: b.input || {} }));
+}
+
+/**
+ * 从 user content blocks 中提取 tool_result
+ */
+function extractToolResults(content) {
+  if (!Array.isArray(content)) return [];
+  return content
+    .filter(b => b.type === 'tool_result')
+    .map(b => {
+      let resultContent;
+      if (typeof b.content === 'string') {
+        resultContent = [{ text: b.content }];
+      } else if (Array.isArray(b.content)) {
+        resultContent = b.content.map(c => {
+          if (typeof c === 'string') return { text: c };
+          if (c.type === 'text') return { text: c.text };
+          return { text: JSON.stringify(c) };
+        });
+      } else {
+        resultContent = [{ text: '' }];
+      }
+      return {
+        toolUseId: b.tool_use_id,
+        content: resultContent,
+        status: b.is_error ? 'error' : 'success',
+      };
+    });
+}
+
+/**
+ * 将 Anthropic 格式的 messages + tools + system 转换为 CodeWhisperer conversationState
+ * 支持完整的工具调用循环
+ */
+export function convertMessages(messages, { modelId, system, tools } = {}) {
+  const validModelId = modelId || undefined;
+  const cwTools = convertTools(tools);
+  const history = [];
+
+  // system → 注入为第一轮 user/assistant 对
+  if (system) {
+    const sysText = typeof system === 'string' ? system : system.map(b => b.text || '').join('\n');
+    if (sysText) {
+      history.push({
+        userInputMessage: {
+          content: sysText, modelId: validModelId, origin: 'AI_EDITOR',
+        },
+      });
+      history.push({ assistantResponseMessage: { content: 'OK' } });
+    }
+  }
+
+  // 遍历 messages，构建 history
+  for (const msg of messages) {
+    if (msg.role === 'user') {
+      const text = extractText(msg.content);
+      const toolResults = extractToolResults(msg.content);
+      const images = extractImages(msg.content);
+
+      if (toolResults.length > 0) {
+        // tool_result 消息：content 为空，结果放在 userInputMessageContext.toolResults
+        history.push({
+          userInputMessage: {
+            content: '',
+            modelId: validModelId,
+            origin: 'AI_EDITOR',
+            userInputMessageContext: { toolResults },
+            ...(images.length > 0 && { images }),
+          },
+        });
+      } else {
+        history.push({
+          userInputMessage: {
+            content: text, modelId: validModelId, origin: 'AI_EDITOR',
+            ...(images.length > 0 && { images }),
+          },
+        });
+      }
+    } else if (msg.role === 'assistant') {
+      const text = extractText(msg.content);
+      const toolUses = extractToolUses(msg.content);
+      history.push({
+        assistantResponseMessage: {
+          content: text,
+          toolUses: toolUses.length > 0 ? toolUses : undefined,
+        },
+      });
+    }
+    // system 已在上面处理
+  }
+
+  // 确保 history 以 user→assistant 交替，末尾是 user
+  // 如果末尾是 assistant（不应该发生），追加空 user
+  const last = history.at(-1);
+  if (last?.assistantResponseMessage) {
+    history.push({
+      userInputMessage: { content: 'Continue.', modelId: validModelId, origin: 'AI_EDITOR' },
+    });
+  }
+
+  const currentMessage = history.at(-1);
+  // 将 tools 注入到 currentMessage
+  if (cwTools && currentMessage?.userInputMessage) {
+    currentMessage.userInputMessage.userInputMessageContext = {
+      ...currentMessage.userInputMessage.userInputMessageContext,
+      tools: cwTools,
+    };
+  }
+
+  return {
+    conversationId: crypto.randomUUID(),
+    currentMessage,
+    history: history.slice(0, -1),
+    chatTriggerType: 'MANUAL',
+  };
+}
+
+// ============================================================
+// 流式调用，返回 text + tool_use 事件
+// ============================================================
+
+/**
+ * 流式调用 Q Developer，yield text 和 tool_use 事件
+ * Claude Code 需要完整的 tool_use 块来驱动工具循环
+ */
+export async function* chatStream(client, { messages, system, tools, profileArn, modelId } = {}) {
+  const conversationState = convertMessages(messages, { modelId, system, tools });
+  const command = new GenerateAssistantResponseCommand({
+    conversationState,
+    profileArn,
+  });
+
+  const response = await client.send(command);
+  if (!response.generateAssistantResponseResponse) {
+    throw new Error('Empty response from Q Developer');
+  }
+
+  // 跟踪当前的 tool_use 状态
+  const activeTools = new Map(); // toolUseId → { name, inputChunks }
+  // 收集统计信息，流结束后汇总输出
+  const stats = {};
+  let meteringUsage = 0;
+
+  for await (const event of response.generateAssistantResponseResponse) {
+    // 文本内容
+    if (event.assistantResponseEvent?.content) {
+      yield {
+        type: 'content',
+        content: event.assistantResponseEvent.content,
+        modelId: event.assistantResponseEvent.modelId,
+      };
+    }
+
+    // thinking/reasoning 内容
+    if (event.reasoningContentEvent) {
+      if (event.reasoningContentEvent.text) {
+        yield { type: 'thinking', text: event.reasoningContentEvent.text };
+      }
+      if (event.reasoningContentEvent.signature) {
+        yield { type: 'thinking_signature', signature: event.reasoningContentEvent.signature };
+      }
+    }
+
+    // 计费/用量事件
+    if (event.meteringEvent) {
+      const m = event.meteringEvent;
+      stats.metering = `${m.usage?.toFixed(4) ?? '?'} ${m.unitPlural || m.unit || 'units'}`;
+      if (typeof m.usage === 'number') meteringUsage = m.usage;
+    }
+
+    // 代码引用/许可证事件
+    if (event.codeReferenceEvent) {
+      stats.codeRef = event.codeReferenceEvent;
+    }
+
+    // 上下文使用率事件
+    if (event.contextUsageEvent) {
+      stats.context = `${(event.contextUsageEvent.contextUsagePercentage ?? 0).toFixed(2)}%`;
+    }
+
+    // token 用量事件
+    if (event.metadataEvent?.tokenUsage) {
+      const t = event.metadataEvent.tokenUsage;
+      const parts = [`in=${t.uncachedInputTokens ?? 0}`, `out=${t.outputTokens ?? 0}`];
+      if (t.cacheReadInputTokens) parts.push(`cache_read=${t.cacheReadInputTokens}`);
+      if (t.cacheWriteInputTokens) parts.push(`cache_write=${t.cacheWriteInputTokens}`);
+      parts.push(`total=${t.totalTokens ?? 0}`);
+      stats.tokens = parts.join(' ');
+    }
+
+    // 无效状态事件（错误）
+    if (event.invalidStateEvent) {
+      stats.invalid = `${event.invalidStateEvent.reason}: ${event.invalidStateEvent.message}`;
+    }
+
+    // 补充链接事件
+    if (event.supplementaryWebLinksEvent?.supplementaryWebLinks?.length) {
+      const links = event.supplementaryWebLinksEvent.supplementaryWebLinks;
+      stats.links = `${links.length} ref(s): ${links.map(l => l.url || l.title).join(', ')}`;
+    }
+
+    // 工具调用事件
+    if (event.toolUseEvent) {
+      const { toolUseId, name, input, stop } = event.toolUseEvent;
+
+      if (toolUseId && name && !activeTools.has(toolUseId)) {
+        // 新工具调用开始
+        activeTools.set(toolUseId, { name, inputChunks: [] });
+        yield { type: 'tool_use_start', toolUseId, name };
+      }
+
+      // 累积 input 片段
+      if (toolUseId && input) {
+        const tool = activeTools.get(toolUseId);
+        if (tool) tool.inputChunks.push(input);
+      }
+
+      // 工具调用结束
+      if (stop) {
+        for (const [id, tool] of activeTools) {
+          // 合并 input 片段并解析
+          let parsedInput = {};
+          const raw = tool.inputChunks.join('');
+          if (raw) {
+            try { parsedInput = JSON.parse(raw); } catch { parsedInput = { raw }; }
+          }
+          yield { type: 'tool_use_end', toolUseId: id, name: tool.name, input: parsedInput };
+        }
+        activeTools.clear();
+      }
+    }
+  }
+
+  // 如果流结束时还有未关闭的工具调用，强制关闭
+  for (const [id, tool] of activeTools) {
+    let parsedInput = {};
+    const raw = tool.inputChunks.join('');
+    if (raw) {
+      try { parsedInput = JSON.parse(raw); } catch { parsedInput = { raw }; }
+    }
+    yield { type: 'tool_use_end', toolUseId: id, name: tool.name, input: parsedInput };
+  }
+
+  // 汇总统计信息
+  yield { type: 'summary', stats, meteringUsage };
+}
+
+/**
+ * 非流式调用
+ */
+export async function chat(client, { messages, system, tools, profileArn, modelId } = {}) {
+  const content = [];
+  let usedModelId;
+  let thinkingText = '';
+  let thinkingSignature;
+  let stats;
+  let meteringUsage = 0;
+
+  for await (const event of chatStream(client, { messages, system, tools, profileArn, modelId })) {
+    if (event.type === 'thinking') {
+      thinkingText += event.text;
+    } else if (event.type === 'thinking_signature') {
+      thinkingSignature = event.signature;
+    } else if (event.type === 'content') {
+      // 如果有累积的 thinking，先输出 thinking block
+      if (thinkingText && !content.some(b => b.type === 'thinking')) {
+        content.push({ type: 'thinking', thinking: thinkingText, signature: thinkingSignature || '' });
+      }
+      if (!content.length || content.at(-1).type !== 'text') {
+        content.push({ type: 'text', text: '' });
+      }
+      content.at(-1).text += event.content;
+      usedModelId = event.modelId;
+    } else if (event.type === 'tool_use_end') {
+      // 如果有累积的 thinking，先输出
+      if (thinkingText && !content.some(b => b.type === 'thinking')) {
+        content.push({ type: 'thinking', thinking: thinkingText, signature: thinkingSignature || '' });
+      }
+      content.push({ type: 'tool_use', id: event.toolUseId, name: event.name, input: event.input });
+    } else if (event.type === 'summary') {
+      stats = event.stats;
+      meteringUsage = event.meteringUsage;
+    }
+  }
+
+  // 如果只有 thinking 没有其他内容
+  if (thinkingText && !content.some(b => b.type === 'thinking')) {
+    content.unshift({ type: 'thinking', thinking: thinkingText, signature: thinkingSignature || '' });
+  }
+
+  const hasToolUse = content.some(b => b.type === 'tool_use');
+  return { content, stopReason: hasToolUse ? 'tool_use' : 'end_turn', modelId: usedModelId, stats, meteringUsage };
+}
+
+// ============================================================
+// ListAvailableModels
+// ============================================================
+
+export async function listAvailableModels(accessToken, { profileArn, authMethod, provider, machineId } = {}) {
+  const arnRegion = regionFromArn(profileArn);
+  const region = arnRegion || DEFAULT_REGION;
+  const endpoint = endpointForRegion(region);
+
+  const params = new URLSearchParams({ origin: 'AI_EDITOR' });
+  if (profileArn) params.set('profileArn', profileArn);
+
+  const headers = {
+    'Authorization': `Bearer ${accessToken}`,
+    'User-Agent': buildUserAgent(machineId),
+    'x-amzn-codewhisperer-optout': 'true',
+  };
+  if (authMethod === 'external_idp') headers['TokenType'] = 'EXTERNAL_IDP';
+  if (provider === 'Internal') headers['redirect-for-internal'] = 'true';
+
+  const allModels = [];
+  let defaultModel = null;
+  let nextToken;
+
+  do {
+    if (nextToken) params.set('nextToken', nextToken);
+    const url = `${endpoint}/ListAvailableModels?${params}`;
+    const res = await fetch(url, { headers });
+    if (!res.ok) {
+      const body = await res.text();
+      throw new Error(`ListAvailableModels failed (${res.status}): ${body}`);
+    }
+    const data = await res.json();
+    allModels.push(...(data.models || []));
+    if (data.defaultModel && !defaultModel) defaultModel = data.defaultModel;
+    nextToken = data.nextToken;
+  } while (nextToken);
+
+  return { models: allModels, defaultModel };
+}