Add Atlas Cloud as a built-in provider (#404)

## Summary
- add Atlas Cloud as a built-in OpenAI-compatible provider with
`https://api.atlascloud.ai/v1`, `ATLASCLOUD_API_KEY`, and
`qwen/qwen3.5-flash` as the default model
- wire Atlas Cloud through provider validation, key-prefix detection,
onboarding IPC parsing, provider shortlist, public base-url fallback,
and provider error help URLs
- update the existing English/Chinese provider lists only; no sponsor
blocks, logos, credits, or partner copy were added

## Validation
- `vitest run packages/shared/src/config.test.ts
packages/shared/src/base-url.test.ts
packages/providers/src/index.test.ts
packages/providers/src/validate.test.ts packages/core/src/errors.test.ts
apps/desktop/src/main/onboarding-ipc.test.ts` (228 tests passed)
- `biome check` on changed source/test/README files
- `tsc --noEmit` for `packages/shared`, `packages/providers`,
`packages/core`, and `apps/desktop/tsconfig.node.json`
- live Atlas Cloud `/v1/models` check returned 118 models, including
`qwen/qwen3.5-flash` and `deepseek-ai/deepseek-v4-pro`
- commit hook ran full-repo `biome check .` successfully

Note: the repository pre-push hook runs the full test suite. In this
local environment, desktop suites that import Electron fail because
Electron's install script is blocked by pnpm's approve-builds policy
(`Electron failed to install correctly`). The targeted
provider/onboarding tests and typechecks above pass.

---------

Co-authored-by: binyangzhu000-sudo <224954946+binyangzhu000-sudo@users.noreply.github.com>
Co-authored-by: Qihan <3045019481@qq.com>

Author: binyangzhu000-sudo

.changeset/atlascloud-provider.md

@@ -0,0 +1,8 @@
+---
+"@open-codesign/shared": patch
+"@open-codesign/providers": patch
+"@open-codesign/core": patch
+"@open-codesign/desktop": patch
+---
+
+Add Atlas Cloud as a built-in OpenAI-compatible provider for onboarding, validation, model discovery, and provider selection.

README.md

@@ -226,7 +226,7 @@ After each stable tag push, CI syncs SHAs back into `packaging/` and publishes d
 On first launch, Open CoDesign opens the Settings page. Pick the path that matches how you already use models:
 
 - **ChatGPT subscription** — sign in with ChatGPT to use Codex models without pasting an API key.
-- **API key** — paste Anthropic (`sk-ant-...`), OpenAI (`sk-...`), Google Gemini, OpenRouter, SiliconFlow, DeepSeek, or another supported provider key.
+- **API key** — paste Anthropic (`sk-ant-...`), OpenAI (`sk-...`), Atlas Cloud, Google Gemini, OpenRouter, SiliconFlow, DeepSeek, or another supported provider key.
 - **Local / keyless** — use Ollama or an IP-allowlisted OpenAI-compatible gateway.
 
 For a custom endpoint that accepts requests without an API key, select **No API key required** in the custom-provider form. Choose the endpoint's wire protocol (including **OpenAI Responses** when appropriate), enter its base URL and default model, and leave the key empty. Local/private endpoint testing still requires the separate network confirmation. Automatic model discovery never sends API keys; for authenticated endpoints, enter a key and click **Test connection**. Switching an existing provider to keyless removes its stored key on save; switching back requires a stored or newly entered key.
@@ -278,7 +278,7 @@ Contract tests check loaded instructions, supported source examples, preference
 ## What you get
 
 ### Models and providers
-- **Unified provider model** — Anthropic, OpenAI, Gemini, DeepSeek, OpenRouter, SiliconFlow, local Ollama, or any OpenAI-compatible relay; keyless (IP-allowlisted) proxies supported
+- **Unified provider model** — Anthropic, OpenAI, Atlas Cloud, Gemini, DeepSeek, OpenRouter, SiliconFlow, local Ollama, or any OpenAI-compatible relay; keyless (IP-allowlisted) proxies supported
 - **One-click import and sign-in** — bring Claude Code / Codex API-key provider configs across, or sign in with ChatGPT subscription for Codex models
 - **Dynamic model picker** — every provider exposes its real model catalogue, not a hardcoded shortlist
 

README.zh-CN.md

@@ -188,7 +188,7 @@ scoop install opencoworkai/open-codesign
 首次启动时，Open CoDesign 会直接打开设置页。按你已有的模型入口选择即可：
 
 - **ChatGPT 订阅登录**：直接登录 ChatGPT，使用 Codex 模型，无需粘贴 API Key。
-- **API Key**：粘贴 Anthropic（`sk-ant-...`）、OpenAI（`sk-...`）、Google Gemini、OpenRouter、SiliconFlow、DeepSeek 或其他支持的 provider key。
+- **API Key**：粘贴 Anthropic（`sk-ant-...`）、OpenAI（`sk-...`）、Atlas Cloud、Google Gemini、OpenRouter、SiliconFlow、DeepSeek 或其他支持的 provider key。
 - **本地 / Keyless**：使用 Ollama，或使用 IP 白名单的 OpenAI 兼容网关。
 
 凭证会保存在 `~/.config/open-codesign/config.toml`，ChatGPT OAuth token 会保存在应用配置目录下的 token store 中。除非你选择的模型入口本身需要联网，请求内容不会额外离开你的机器。
@@ -220,7 +220,7 @@ scoop install opencoworkai/open-codesign
 ## 你能得到什么
 
 ### 模型与提供商
-- **统一的 provider 抽象**：支持 Anthropic、OpenAI、Gemini、DeepSeek、OpenRouter、SiliconFlow、本地 Ollama，以及任意 OpenAI-compatible relay；同时支持无 key 的 IP 白名单代理
+- **统一的 provider 抽象**：支持 Anthropic、OpenAI、Atlas Cloud、Gemini、DeepSeek、OpenRouter、SiliconFlow、本地 Ollama，以及任意 OpenAI-compatible relay；同时支持无 key 的 IP 白名单代理
 - **一键导入和登录**：Claude Code / Codex 的 API key provider 配置可以直接带进来，也可以用 ChatGPT 订阅登录使用 Codex 模型
 - **动态模型选择器**：每个 provider 都会展示真实模型列表，而不是一小撮写死的选项
 

apps/desktop/src/main/onboarding-ipc.test.ts

@@ -903,6 +903,25 @@ describe('registerOnboardingIpc — validate-key passes baseUrl to pingProvider'
     expect(pingProvider).toHaveBeenCalledWith('anthropic', 'sk-ant-test', undefined);
   });
 
+  it('forwards Atlas Cloud validation to the provider validator', async () => {
+    const { pingProvider } = await import('@open-codesign/providers');
+    vi.mocked(pingProvider).mockClear();
+    const handler = handlers.get('onboarding:validate-key');
+    expect(handler).toBeDefined();
+
+    await handler?.({} as unknown, {
+      provider: 'atlascloud',
+      apiKey: 'apikey-test',
+      baseUrl: 'https://api.atlascloud.ai/v1',
+    });
+
+    expect(pingProvider).toHaveBeenCalledWith(
+      'atlascloud',
+      'apikey-test',
+      'https://api.atlascloud.ai/v1',
+    );
+  });
+
   it('allows explicitly keyless Ollama validation with an empty apiKey', async () => {
     const { pingProvider } = await import('@open-codesign/providers');
     vi.mocked(pingProvider).mockClear();

apps/desktop/src/main/onboarding/provider-parsers.ts

@@ -222,7 +222,7 @@ export function parseValidateKey(raw: unknown): ValidateKeyInput {
   const providerId = provider.trim();
   if (!isSupportedOnboardingProvider(providerId)) {
     throw new CodesignError(
-      `Provider "${providerId}" is not supported. Only anthropic, openai, openrouter, ollama.`,
+      `Provider "${providerId}" is not supported. Only anthropic, openai, atlascloud, openrouter, ollama.`,
       ERROR_CODES.PROVIDER_NOT_SUPPORTED,
     );
   }

packages/core/src/agent.ts

@@ -271,6 +271,7 @@ function supportsImageInput(wire: WireApi | undefined, modelId: string): boolean
 const BUILTIN_PUBLIC_BASE_URLS: Record<string, string> = {
   anthropic: 'https://api.anthropic.com',
   openai: 'https://api.openai.com/v1',
+  atlascloud: 'https://api.atlascloud.ai/v1',
   openrouter: 'https://openrouter.ai/api/v1',
 };
 

packages/core/src/errors.test.ts

@@ -30,6 +30,13 @@ describe('rewriteUpstreamMessage', () => {
     expect(result.message).toContain('openrouter.ai/settings/keys');
   });
 
+  it('rewrites leaked openai URL to Atlas Cloud URL', () => {
+    const result = rewriteUpstreamMessage(LEAKED, 'atlascloud', 401);
+    expect(result.rewritten).toBe(true);
+    expect(result.message).not.toContain('openai.com');
+    expect(result.message).toContain('atlascloud.ai');
+  });
+
   it('rewrites to deepseek URL even though it is not in the typed enum', () => {
     const result = rewriteUpstreamMessage(LEAKED, 'deepseek', 401);
     expect(result.message).toContain('platform.deepseek.com/api_keys');

packages/core/src/errors.ts

@@ -21,6 +21,7 @@ import { CodesignError, ERROR_CODES } from '@open-codesign/shared';
 
 export const PROVIDER_KEY_HELP_URL: Partial<Record<ProviderId, string>> = {
   openai: 'https://platform.openai.com/account/api-keys',
+  atlascloud: 'https://atlascloud.ai/',
   anthropic: 'https://console.anthropic.com/settings/keys',
   openrouter: 'https://openrouter.ai/settings/keys',
   google: 'https://aistudio.google.com/app/apikey',

packages/providers/src/index.test.ts

@@ -9,7 +9,7 @@ vi.mock('@mariozechner/pi-ai', () => ({
   completeSimple: (...args: unknown[]) => completeSimpleMock(...args),
 }));
 
-import { complete, inferReasoning } from './index';
+import { complete, detectProviderFromKey, inferReasoning } from './index';
 
 const MODEL: ModelRef = { provider: 'openai', modelId: 'gpt-4o' };
 
@@ -18,6 +18,12 @@ afterEach(() => {
   completeSimpleMock.mockReset();
 });
 
+describe('detectProviderFromKey', () => {
+  it('recognizes Atlas Cloud API keys', () => {
+    expect(detectProviderFromKey(' apikey-test ')).toBe('atlascloud');
+  });
+});
+
 describe('complete', () => {
   it.each([
     [undefined, 'low'],

packages/providers/src/index.ts

@@ -592,6 +592,7 @@ export function detectProviderFromKey(key: string): ModelRef['provider'] | null
   const trimmed = key.trim();
   if (trimmed.startsWith('sk-ant-')) return 'anthropic';
   if (trimmed.startsWith('sk-or-')) return 'openrouter';
+  if (trimmed.startsWith('apikey-')) return 'atlascloud';
   if (trimmed.startsWith('sk-')) return 'openai';
   if (trimmed.startsWith('AIza')) return 'google';
   if (trimmed.startsWith('xai-')) return 'xai';