ai-zero-token 2.0.1 → 2.0.3

package/admin-ui/dist/assets/docs-DtO-AOWU.js

@@ -0,0 +1,1735 @@
+import{a as e,n as t,r as n,t as r}from"./jsx-runtime-DqpGtLhh.js";import{t as i}from"./server-BrjJPb9D.js";import{b as a,f as o,g as s,n as c,v as l,y as u}from"./index-DRe-tByu.js";var d=t(`arrow-right`,[[`path`,{d:`M5 12h14`,key:`1ays0h`}],[`path`,{d:`m12 5 7 7-7 7`,key:`xquz4c`}]]),f=`# AI-Zero-Token Local Gateway Skill
+
+## Purpose
+
+This skill tells the LLM how to integrate a webpage, local tool, or Vibe Coding project with the local AI-Zero-Token gateway.
+
+AI-Zero-Token is a local-first personal AI gateway. It exposes OpenAI-style local HTTP APIs so that a frontend page, script, or local automation tool can call available ChatGPT / Codex account capabilities through a local endpoint.
+
+Use this skill when the user wants to build:
+
+- AI chat assistant
+- text generation tool
+- copywriting generator
+- prompt generator
+- image generation tool
+- image editing / image-to-image tool
+- poster generator
+- product image generator
+- social media cover generator
+- icon or logo inspiration generator
+- UI inspiration generator
+- Vibe Coding webpage that needs AI capability
+- local automation workflow that needs LLM or image generation
+
+---
+
+## Local Gateway Assumptions
+
+Assume the user has already started AI-Zero-Token locally.
+
+Default local management page:
+
+\`\`\`text
+http://127.0.0.1:8787
+\`\`\`
+
+OpenAI-compatible Base URL:
+
+\`\`\`text
+http://127.0.0.1:8787/v1
+\`\`\`
+
+When using OpenAI-compatible SDKs or API clients, set:
+
+\`\`\`text
+baseURL = "http://127.0.0.1:8787/v1"
+apiKey = "local"
+\`\`\`
+
+The API key can be any non-empty placeholder string because the real authentication is handled by the local AI-Zero-Token gateway.
+
+Do not ask the user to provide a real OpenAI API key unless the user explicitly wants to use the official OpenAI API instead of the local gateway.
+
+---
+
+## Startup Command Reference
+
+If the generated project includes user instructions, mention the user can start AI-Zero-Token with:
+
+\`\`\`bash
+npx ai-zero-token start
+\`\`\`
+
+or install globally:
+
+\`\`\`bash
+npm install -g ai-zero-token
+azt start
+\`\`\`
+
+After startup, the local management page is usually available at:
+
+\`\`\`text
+http://127.0.0.1:8787
+\`\`\`
+
+The OpenAI-compatible API Base URL is:
+
+\`\`\`text
+http://127.0.0.1:8787/v1
+\`\`\`
+
+If global npm installation fails with \`EACCES permission denied\`, recommend using:
+
+\`\`\`bash
+npx ai-zero-token start
+\`\`\`
+
+---
+
+## Available Interfaces
+
+### 1. List Models
+
+Use this endpoint when the app needs to show available models.
+
+\`\`\`http
+GET http://127.0.0.1:8787/v1/models
+\`\`\`
+
+Example JavaScript:
+
+\`\`\`js
+async function listModels() {
+  const res = await fetch("http://127.0.0.1:8787/v1/models", {
+    method: "GET",
+    headers: {
+      "Authorization": "Bearer local"
+    }
+  });
+
+  if (!res.ok) {
+    throw new Error(\`Failed to list models: \${res.status}\`);
+  }
+
+  return await res.json();
+}
+\`\`\`
+
+---
+
+### 2. Text / Reasoning / Coding Generation
+
+Preferred endpoint:
+
+\`\`\`http
+POST http://127.0.0.1:8787/v1/responses
+\`\`\`
+
+Use this endpoint for:
+
+- AI assistant replies
+- text generation
+- copywriting
+- prompt optimization
+- coding assistance
+- reasoning tasks
+- content rewriting
+- document drafting
+
+Default recommended model:
+
+\`\`\`text
+gpt-5.5
+\`\`\`
+
+If \`gpt-5.5\` is not available, allow the user or UI to switch models.
+
+Example request:
+
+\`\`\`js
+async function generateText(input) {
+  const res = await fetch("http://127.0.0.1:8787/v1/responses", {
+    method: "POST",
+    headers: {
+      "Content-Type": "application/json",
+      "Authorization": "Bearer local"
+    },
+    body: JSON.stringify({
+      model: "gpt-5.5",
+      input
+    })
+  });
+
+  if (!res.ok) {
+    const errText = await res.text();
+    throw new Error(\`Text generation failed: \${res.status} \${errText}\`);
+  }
+
+  const data = await res.json();
+
+  return (
+    data.output_text ||
+    data.text ||
+    data.content ||
+    data.choices?.[0]?.message?.content ||
+    JSON.stringify(data, null, 2)
+  );
+}
+\`\`\`
+
+When building a frontend app, show loading state while the request is pending and display a readable error message if the request fails.
+
+---
+
+### 3. Chat Completions Compatibility
+
+Use this endpoint when the user asks for an OpenAI-compatible chat app or when a template already uses \`/v1/chat/completions\`.
+
+\`\`\`http
+POST http://127.0.0.1:8787/v1/chat/completions
+\`\`\`
+
+Example request:
+
+\`\`\`js
+async function chat(messages) {
+  const res = await fetch("http://127.0.0.1:8787/v1/chat/completions", {
+    method: "POST",
+    headers: {
+      "Content-Type": "application/json",
+      "Authorization": "Bearer local"
+    },
+    body: JSON.stringify({
+      model: "gpt-5.5",
+      messages
+    })
+  });
+
+  if (!res.ok) {
+    const errText = await res.text();
+    throw new Error(\`Chat completion failed: \${res.status} \${errText}\`);
+  }
+
+  const data = await res.json();
+
+  return (
+    data.choices?.[0]?.message?.content ||
+    data.output_text ||
+    JSON.stringify(data, null, 2)
+  );
+}
+\`\`\`
+
+Use this for chat UI components, OpenAI-compatible templates, or existing chat SDK style code.
+
+---
+
+### 4. Image Generation
+
+Use this endpoint when the user wants image generation.
+
+\`\`\`http
+POST http://127.0.0.1:8787/v1/images/generations
+\`\`\`
+
+Use this endpoint for:
+
+- poster generation
+- product image generation
+- social media image generation
+- cover image generation
+- icon inspiration
+- logo inspiration
+- UI visual inspiration
+- marketing image generation
+- illustration generation
+
+Default recommended image model:
+
+\`\`\`text
+gpt-image-2
+\`\`\`
+
+Example request:
+
+\`\`\`js
+async function generateImage(prompt, size = "1024x1024") {
+  const res = await fetch("http://127.0.0.1:8787/v1/images/generations", {
+    method: "POST",
+    headers: {
+      "Content-Type": "application/json",
+      "Authorization": "Bearer local"
+    },
+    body: JSON.stringify({
+      model: "gpt-image-2",
+      prompt,
+      size,
+      n: 1
+    })
+  });
+
+  if (!res.ok) {
+    const errText = await res.text();
+    throw new Error(\`Image generation failed: \${res.status} \${errText}\`);
+  }
+
+  const data = await res.json();
+
+  const image =
+    data.data?.[0]?.url ||
+    data.data?.[0]?.b64_json ||
+    data.output?.[0]?.url ||
+    data.url ||
+    data.image;
+
+  return {
+    raw: data,
+    image
+  };
+}
+\`\`\`
+
+When displaying generated images:
+
+- If the response contains a URL, use it directly as the \`src\`.
+- If the response contains \`b64_json\`, convert it to a data URL:
+
+\`\`\`js
+const src = \`data:image/png;base64,\${b64_json}\`;
+\`\`\`
+
+- Always show the raw response in a collapsible debug panel if this is a prototype or developer-facing tool.
+- Show a helpful error if no image URL or base64 data is found.
+
+---
+
+### 5. Image Editing / Image-to-Image
+
+Use this endpoint when the user wants to edit an existing image or use one or more reference images.
+
+\`\`\`http
+POST http://127.0.0.1:8787/v1/images/edits
+\`\`\`
+
+AI-Zero-Token currently supports the JSON version of this endpoint. Use \`images[].image_url\` with either:
+
+- an \`https://...\` image URL
+- a \`data:image/...;base64,...\` data URL
+- a bare base64 string, which the gateway treats as a PNG data URL
+
+Do not use \`multipart/form-data\`, \`mask\`, or \`file_id\` unless the user explicitly says their installed AI-Zero-Token version supports them.
+
+Example request:
+
+\`\`\`js
+async function editImage(prompt, imageUrlOrBase64, size = "1024x1024") {
+  const res = await fetch("http://127.0.0.1:8787/v1/images/edits", {
+    method: "POST",
+    headers: {
+      "Content-Type": "application/json",
+      "Authorization": "Bearer local"
+    },
+    body: JSON.stringify({
+      model: "gpt-image-2",
+      prompt,
+      images: [
+        {
+          image_url: imageUrlOrBase64
+        }
+      ],
+      size,
+      quality: "low",
+      response_format: "b64_json"
+    })
+  });
+
+  if (!res.ok) {
+    const errText = await res.text();
+    throw new Error(\`Image edit failed: \${res.status} \${errText}\`);
+  }
+
+  const data = await res.json();
+  const image = data.data?.[0]?.b64_json || data.data?.[0]?.url || data.image;
+
+  return {
+    raw: data,
+    image
+  };
+}
+\`\`\`
+
+When building a frontend for image editing:
+
+- Let the user paste an image URL.
+- Optionally support local file upload by converting the file to a base64 data URL in the browser, then send it as \`images[].image_url\`.
+- Show the edited image using \`data:image/png;base64,...\` when the response contains \`b64_json\`.
+- Mention that mask/inpainting is not supported by this JSON integration yet.
+
+---
+
+## Frontend Integration Rules
+
+When generating a web app that uses AI-Zero-Token:
+
+1. Use \`http://127.0.0.1:8787/v1\` as the default Base URL.
+2. Use \`local\` as the default API Key placeholder.
+3. Do not hardcode a real OpenAI API key.
+4. Provide editable fields for:
+   - Base URL
+   - text model name
+   - image model name
+   - prompt / input
+5. Use \`gpt-5.5\` as the default text model.
+6. Use \`gpt-image-2\` as the default image model.
+7. Add loading states.
+8. Add error display.
+9. Add a clear result area.
+10. Add a copy button for text outputs.
+11. Add an image preview area for image outputs.
+12. For image editing tools, accept either pasted image URLs or local files converted to base64 data URLs.
+13. Add a download button for generated or edited images when possible.
+14. Keep all calls client-side for local prototype projects unless the user asks for a backend.
+15. If the browser reports a CORS error, tell the user to check whether AI-Zero-Token is running and whether the gateway allows browser requests.
+
+---
+
+## Recommended UI Patterns
+
+For a simple Vibe Coding webpage, create:
+
+- A settings panel:
+  - Base URL
+  - Text model
+  - Image model
+- A text generation panel:
+  - textarea for input
+  - button: Generate Text
+  - output area
+- An image generation panel:
+  - textarea for prompt
+  - size selector
+  - button: Generate Image
+  - image preview area
+- For image editing tools:
+  - image URL input or file picker
+  - textarea for edit instruction
+  - button: Edit Image
+  - before / after image preview
+- A debug panel:
+  - raw JSON response
+  - error message
+
+Use a clean layout that is easy for non-developers to understand.
+
+---
+
+## Default Models
+
+Use these defaults unless the user says otherwise:
+
+\`\`\`text
+Text model: gpt-5.5
+Image model: gpt-image-2
+\`\`\`
+
+If a model call fails, suggest:
+
+1. Check whether AI-Zero-Token is running.
+2. Open \`http://127.0.0.1:8787\`.
+3. Check whether the account is logged in.
+4. Check whether the current account supports the requested model.
+5. Try calling \`GET /v1/models\`.
+
+---
+
+## Example Prompt for Building a Vibe Coding Tool
+
+When the user asks to build a webpage using this skill, generate something like:
+
+\`\`\`text
+Build a local AI tool webpage that connects to AI-Zero-Token.
+
+Use:
+Base URL: http://127.0.0.1:8787/v1
+API Key: local
+
+Features:
+1. A text input area for user requirements.
+2. A "Generate Text" button that calls POST /v1/responses with model gpt-5.5.
+3. A "Generate Image" button that calls POST /v1/images/generations with model gpt-image-2.
+4. If the tool needs image editing, call POST /v1/images/edits with images[].image_url.
+5. Show loading states.
+6. Show text output in a result panel.
+7. Show generated or edited image in an image preview panel.
+8. Provide a copy button for text.
+9. Provide a download button for image if possible.
+10. Show readable errors if the local gateway is not running.
+11. Keep the UI simple and friendly for designers.
+\`\`\`
+
+---
+
+## Common Error Handling
+
+### Gateway not running
+
+If request fails with network error, show:
+
+\`\`\`text
+AI-Zero-Token 本地网关未连接，请先运行 npx ai-zero-token start 或 azt start，然后访问 http://127.0.0.1:8787 检查状态。
+\`\`\`
+
+### CORS error
+
+If browser blocks the request, show:
+
+\`\`\`text
+浏览器跨域请求失败，请确认 AI-Zero-Token 网关已开启浏览器访问支持，或改用本地代理 / 后端转发。
+\`\`\`
+
+### Model unavailable
+
+If the model is unavailable, show:
+
+\`\`\`text
+当前账号可能不支持该模型，请打开 AI-Zero-Token 管理页检查账号状态和可用模型。
+\`\`\`
+
+### Image result not found
+
+If the image response does not contain \`url\` or \`b64_json\`, show:
+
+\`\`\`text
+接口已返回，但没有找到可展示的图片地址或 base64 数据，请查看调试面板中的原始返回结果。
+\`\`\`
+
+### Image edit unsupported input
+
+If image editing fails because of unsupported input, show:
+
+\`\`\`text
+当前 AI-Zero-Token 的图片编辑接口只支持 JSON 请求中的 image_url。请使用图片 URL，或把本地图片转换成 base64 data URL 后再提交。
+\`\`\`
+
+### npm global install permission error
+
+If the user gets an error like \`EACCES permission denied\` during global installation, suggest:
+
+\`\`\`bash
+npx ai-zero-token start
+\`\`\`
+
+or explain that global npm installation may require changing the npm global prefix or using a Node version manager.
+
+---
+
+## Important Boundaries
+
+AI-Zero-Token is best for:
+
+- personal experiments
+- local prototypes
+- Vibe Coding projects
+- designer-facing demos
+- learning local gateway integration
+- low-cost local workflow validation
+
+AI-Zero-Token management page can also manage local account state, including:
+
+- login and import multiple Codex accounts
+- switch the active gateway account
+- export selected accounts
+- apply a saved account to local Codex by updating \`~/.codex/auth.json\` with an automatic backup
+
+AI-Zero-Token should not be described as:
+
+- a production-grade API gateway
+- an official OpenAI API replacement
+- a commercial proxy platform
+- a multi-user billing platform
+
+For production systems, recommend official APIs or a properly managed backend gateway.
+
+---
+
+## Short Usage Instruction for Designers
+
+Designers do not need to understand this skill document.
+
+They can give this file to a Vibe Coding LLM and say:
+
+\`\`\`text
+请根据这个 Skill，帮我做一个 AI 生图网页。
+要求：
+1. 输入一句图片需求；
+2. 点击按钮调用本地 AI-Zero-Token；
+3. 使用 gpt-image-2 生成图片；
+4. 图片生成后展示在页面中；
+5. 页面要好看，适合设计师使用。
+\`\`\`
+
+Or:
+
+\`\`\`text
+请根据这个 Skill，帮我做一个 AI 文案 + 生图工具。
+左侧输入需求，右侧显示 GPT-5.5 生成的文案和 GPT-Image-2 生成的图片。
+\`\`\`
+`,p=e(n(),1),m=r();function h({title:e,description:t,code:n,onCopy:r}){return(0,m.jsxs)(`section`,{className:`docs-snippet`,children:[(0,m.jsxs)(`div`,{className:`docs-snippet-head`,children:[(0,m.jsxs)(`div`,{children:[(0,m.jsx)(`strong`,{children:e}),(0,m.jsx)(`p`,{children:t})]}),(0,m.jsx)(`button`,{className:`btn-secondary icon-only`,type:`button`,onClick:r,title:`复制代码`,children:(0,m.jsx)(a,{size:16})})]}),(0,m.jsx)(`pre`,{children:(0,m.jsx)(`code`,{children:n})})]})}function g({config:e,onRoute:t,copyBaseUrl:n,setStatus:r}){let[g,_]=(0,p.useState)(`quick-start`),v=e?.baseUrl||`http://127.0.0.1:8787/v1`,y=`local`,b=`azt start`,x=f.split(/\r?\n/).length,S=(0,p.useMemo)(()=>({lines:f.split(/\r?\n/).length,codeBlocks:(`# AI-Zero-Token Local Gateway Skill
+
+## Purpose
+
+This skill tells the LLM how to integrate a webpage, local tool, or Vibe Coding project with the local AI-Zero-Token gateway.
+
+AI-Zero-Token is a local-first personal AI gateway. It exposes OpenAI-style local HTTP APIs so that a frontend page, script, or local automation tool can call available ChatGPT / Codex account capabilities through a local endpoint.
+
+Use this skill when the user wants to build:
+
+- AI chat assistant
+- text generation tool
+- copywriting generator
+- prompt generator
+- image generation tool
+- image editing / image-to-image tool
+- poster generator
+- product image generator
+- social media cover generator
+- icon or logo inspiration generator
+- UI inspiration generator
+- Vibe Coding webpage that needs AI capability
+- local automation workflow that needs LLM or image generation
+
+---
+
+## Local Gateway Assumptions
+
+Assume the user has already started AI-Zero-Token locally.
+
+Default local management page:
+
+\`\`\`text
+http://127.0.0.1:8787
+\`\`\`
+
+OpenAI-compatible Base URL:
+
+\`\`\`text
+http://127.0.0.1:8787/v1
+\`\`\`
+
+When using OpenAI-compatible SDKs or API clients, set:
+
+\`\`\`text
+baseURL = "http://127.0.0.1:8787/v1"
+apiKey = "local"
+\`\`\`
+
+The API key can be any non-empty placeholder string because the real authentication is handled by the local AI-Zero-Token gateway.
+
+Do not ask the user to provide a real OpenAI API key unless the user explicitly wants to use the official OpenAI API instead of the local gateway.
+
+---
+
+## Startup Command Reference
+
+If the generated project includes user instructions, mention the user can start AI-Zero-Token with:
+
+\`\`\`bash
+npx ai-zero-token start
+\`\`\`
+
+or install globally:
+
+\`\`\`bash
+npm install -g ai-zero-token
+azt start
+\`\`\`
+
+After startup, the local management page is usually available at:
+
+\`\`\`text
+http://127.0.0.1:8787
+\`\`\`
+
+The OpenAI-compatible API Base URL is:
+
+\`\`\`text
+http://127.0.0.1:8787/v1
+\`\`\`
+
+If global npm installation fails with \`EACCES permission denied\`, recommend using:
+
+\`\`\`bash
+npx ai-zero-token start
+\`\`\`
+
+---
+
+## Available Interfaces
+
+### 1. List Models
+
+Use this endpoint when the app needs to show available models.
+
+\`\`\`http
+GET http://127.0.0.1:8787/v1/models
+\`\`\`
+
+Example JavaScript:
+
+\`\`\`js
+async function listModels() {
+  const res = await fetch("http://127.0.0.1:8787/v1/models", {
+    method: "GET",
+    headers: {
+      "Authorization": "Bearer local"
+    }
+  });
+
+  if (!res.ok) {
+    throw new Error(\`Failed to list models: \${res.status}\`);
+  }
+
+  return await res.json();
+}
+\`\`\`
+
+---
+
+### 2. Text / Reasoning / Coding Generation
+
+Preferred endpoint:
+
+\`\`\`http
+POST http://127.0.0.1:8787/v1/responses
+\`\`\`
+
+Use this endpoint for:
+
+- AI assistant replies
+- text generation
+- copywriting
+- prompt optimization
+- coding assistance
+- reasoning tasks
+- content rewriting
+- document drafting
+
+Default recommended model:
+
+\`\`\`text
+gpt-5.5
+\`\`\`
+
+If \`gpt-5.5\` is not available, allow the user or UI to switch models.
+
+Example request:
+
+\`\`\`js
+async function generateText(input) {
+  const res = await fetch("http://127.0.0.1:8787/v1/responses", {
+    method: "POST",
+    headers: {
+      "Content-Type": "application/json",
+      "Authorization": "Bearer local"
+    },
+    body: JSON.stringify({
+      model: "gpt-5.5",
+      input
+    })
+  });
+
+  if (!res.ok) {
+    const errText = await res.text();
+    throw new Error(\`Text generation failed: \${res.status} \${errText}\`);
+  }
+
+  const data = await res.json();
+
+  return (
+    data.output_text ||
+    data.text ||
+    data.content ||
+    data.choices?.[0]?.message?.content ||
+    JSON.stringify(data, null, 2)
+  );
+}
+\`\`\`
+
+When building a frontend app, show loading state while the request is pending and display a readable error message if the request fails.
+
+---
+
+### 3. Chat Completions Compatibility
+
+Use this endpoint when the user asks for an OpenAI-compatible chat app or when a template already uses \`/v1/chat/completions\`.
+
+\`\`\`http
+POST http://127.0.0.1:8787/v1/chat/completions
+\`\`\`
+
+Example request:
+
+\`\`\`js
+async function chat(messages) {
+  const res = await fetch("http://127.0.0.1:8787/v1/chat/completions", {
+    method: "POST",
+    headers: {
+      "Content-Type": "application/json",
+      "Authorization": "Bearer local"
+    },
+    body: JSON.stringify({
+      model: "gpt-5.5",
+      messages
+    })
+  });
+
+  if (!res.ok) {
+    const errText = await res.text();
+    throw new Error(\`Chat completion failed: \${res.status} \${errText}\`);
+  }
+
+  const data = await res.json();
+
+  return (
+    data.choices?.[0]?.message?.content ||
+    data.output_text ||
+    JSON.stringify(data, null, 2)
+  );
+}
+\`\`\`
+
+Use this for chat UI components, OpenAI-compatible templates, or existing chat SDK style code.
+
+---
+
+### 4. Image Generation
+
+Use this endpoint when the user wants image generation.
+
+\`\`\`http
+POST http://127.0.0.1:8787/v1/images/generations
+\`\`\`
+
+Use this endpoint for:
+
+- poster generation
+- product image generation
+- social media image generation
+- cover image generation
+- icon inspiration
+- logo inspiration
+- UI visual inspiration
+- marketing image generation
+- illustration generation
+
+Default recommended image model:
+
+\`\`\`text
+gpt-image-2
+\`\`\`
+
+Example request:
+
+\`\`\`js
+async function generateImage(prompt, size = "1024x1024") {
+  const res = await fetch("http://127.0.0.1:8787/v1/images/generations", {
+    method: "POST",
+    headers: {
+      "Content-Type": "application/json",
+      "Authorization": "Bearer local"
+    },
+    body: JSON.stringify({
+      model: "gpt-image-2",
+      prompt,
+      size,
+      n: 1
+    })
+  });
+
+  if (!res.ok) {
+    const errText = await res.text();
+    throw new Error(\`Image generation failed: \${res.status} \${errText}\`);
+  }
+
+  const data = await res.json();
+
+  const image =
+    data.data?.[0]?.url ||
+    data.data?.[0]?.b64_json ||
+    data.output?.[0]?.url ||
+    data.url ||
+    data.image;
+
+  return {
+    raw: data,
+    image
+  };
+}
+\`\`\`
+
+When displaying generated images:
+
+- If the response contains a URL, use it directly as the \`src\`.
+- If the response contains \`b64_json\`, convert it to a data URL:
+
+\`\`\`js
+const src = \`data:image/png;base64,\${b64_json}\`;
+\`\`\`
+
+- Always show the raw response in a collapsible debug panel if this is a prototype or developer-facing tool.
+- Show a helpful error if no image URL or base64 data is found.
+
+---
+
+### 5. Image Editing / Image-to-Image
+
+Use this endpoint when the user wants to edit an existing image or use one or more reference images.
+
+\`\`\`http
+POST http://127.0.0.1:8787/v1/images/edits
+\`\`\`
+
+AI-Zero-Token currently supports the JSON version of this endpoint. Use \`images[].image_url\` with either:
+
+- an \`https://...\` image URL
+- a \`data:image/...;base64,...\` data URL
+- a bare base64 string, which the gateway treats as a PNG data URL
+
+Do not use \`multipart/form-data\`, \`mask\`, or \`file_id\` unless the user explicitly says their installed AI-Zero-Token version supports them.
+
+Example request:
+
+\`\`\`js
+async function editImage(prompt, imageUrlOrBase64, size = "1024x1024") {
+  const res = await fetch("http://127.0.0.1:8787/v1/images/edits", {
+    method: "POST",
+    headers: {
+      "Content-Type": "application/json",
+      "Authorization": "Bearer local"
+    },
+    body: JSON.stringify({
+      model: "gpt-image-2",
+      prompt,
+      images: [
+        {
+          image_url: imageUrlOrBase64
+        }
+      ],
+      size,
+      quality: "low",
+      response_format: "b64_json"
+    })
+  });
+
+  if (!res.ok) {
+    const errText = await res.text();
+    throw new Error(\`Image edit failed: \${res.status} \${errText}\`);
+  }
+
+  const data = await res.json();
+  const image = data.data?.[0]?.b64_json || data.data?.[0]?.url || data.image;
+
+  return {
+    raw: data,
+    image
+  };
+}
+\`\`\`
+
+When building a frontend for image editing:
+
+- Let the user paste an image URL.
+- Optionally support local file upload by converting the file to a base64 data URL in the browser, then send it as \`images[].image_url\`.
+- Show the edited image using \`data:image/png;base64,...\` when the response contains \`b64_json\`.
+- Mention that mask/inpainting is not supported by this JSON integration yet.
+
+---
+
+## Frontend Integration Rules
+
+When generating a web app that uses AI-Zero-Token:
+
+1. Use \`http://127.0.0.1:8787/v1\` as the default Base URL.
+2. Use \`local\` as the default API Key placeholder.
+3. Do not hardcode a real OpenAI API key.
+4. Provide editable fields for:
+   - Base URL
+   - text model name
+   - image model name
+   - prompt / input
+5. Use \`gpt-5.5\` as the default text model.
+6. Use \`gpt-image-2\` as the default image model.
+7. Add loading states.
+8. Add error display.
+9. Add a clear result area.
+10. Add a copy button for text outputs.
+11. Add an image preview area for image outputs.
+12. For image editing tools, accept either pasted image URLs or local files converted to base64 data URLs.
+13. Add a download button for generated or edited images when possible.
+14. Keep all calls client-side for local prototype projects unless the user asks for a backend.
+15. If the browser reports a CORS error, tell the user to check whether AI-Zero-Token is running and whether the gateway allows browser requests.
+
+---
+
+## Recommended UI Patterns
+
+For a simple Vibe Coding webpage, create:
+
+- A settings panel:
+  - Base URL
+  - Text model
+  - Image model
+- A text generation panel:
+  - textarea for input
+  - button: Generate Text
+  - output area
+- An image generation panel:
+  - textarea for prompt
+  - size selector
+  - button: Generate Image
+  - image preview area
+- For image editing tools:
+  - image URL input or file picker
+  - textarea for edit instruction
+  - button: Edit Image
+  - before / after image preview
+- A debug panel:
+  - raw JSON response
+  - error message
+
+Use a clean layout that is easy for non-developers to understand.
+
+---
+
+## Default Models
+
+Use these defaults unless the user says otherwise:
+
+\`\`\`text
+Text model: gpt-5.5
+Image model: gpt-image-2
+\`\`\`
+
+If a model call fails, suggest:
+
+1. Check whether AI-Zero-Token is running.
+2. Open \`http://127.0.0.1:8787\`.
+3. Check whether the account is logged in.
+4. Check whether the current account supports the requested model.
+5. Try calling \`GET /v1/models\`.
+
+---
+
+## Example Prompt for Building a Vibe Coding Tool
+
+When the user asks to build a webpage using this skill, generate something like:
+
+\`\`\`text
+Build a local AI tool webpage that connects to AI-Zero-Token.
+
+Use:
+Base URL: http://127.0.0.1:8787/v1
+API Key: local
+
+Features:
+1. A text input area for user requirements.
+2. A "Generate Text" button that calls POST /v1/responses with model gpt-5.5.
+3. A "Generate Image" button that calls POST /v1/images/generations with model gpt-image-2.
+4. If the tool needs image editing, call POST /v1/images/edits with images[].image_url.
+5. Show loading states.
+6. Show text output in a result panel.
+7. Show generated or edited image in an image preview panel.
+8. Provide a copy button for text.
+9. Provide a download button for image if possible.
+10. Show readable errors if the local gateway is not running.
+11. Keep the UI simple and friendly for designers.
+\`\`\`
+
+---
+
+## Common Error Handling
+
+### Gateway not running
+
+If request fails with network error, show:
+
+\`\`\`text
+AI-Zero-Token 本地网关未连接，请先运行 npx ai-zero-token start 或 azt start，然后访问 http://127.0.0.1:8787 检查状态。
+\`\`\`
+
+### CORS error
+
+If browser blocks the request, show:
+
+\`\`\`text
+浏览器跨域请求失败，请确认 AI-Zero-Token 网关已开启浏览器访问支持，或改用本地代理 / 后端转发。
+\`\`\`
+
+### Model unavailable
+
+If the model is unavailable, show:
+
+\`\`\`text
+当前账号可能不支持该模型，请打开 AI-Zero-Token 管理页检查账号状态和可用模型。
+\`\`\`
+
+### Image result not found
+
+If the image response does not contain \`url\` or \`b64_json\`, show:
+
+\`\`\`text
+接口已返回，但没有找到可展示的图片地址或 base64 数据，请查看调试面板中的原始返回结果。
+\`\`\`
+
+### Image edit unsupported input
+
+If image editing fails because of unsupported input, show:
+
+\`\`\`text
+当前 AI-Zero-Token 的图片编辑接口只支持 JSON 请求中的 image_url。请使用图片 URL，或把本地图片转换成 base64 data URL 后再提交。
+\`\`\`
+
+### npm global install permission error
+
+If the user gets an error like \`EACCES permission denied\` during global installation, suggest:
+
+\`\`\`bash
+npx ai-zero-token start
+\`\`\`
+
+or explain that global npm installation may require changing the npm global prefix or using a Node version manager.
+
+---
+
+## Important Boundaries
+
+AI-Zero-Token is best for:
+
+- personal experiments
+- local prototypes
+- Vibe Coding projects
+- designer-facing demos
+- learning local gateway integration
+- low-cost local workflow validation
+
+AI-Zero-Token management page can also manage local account state, including:
+
+- login and import multiple Codex accounts
+- switch the active gateway account
+- export selected accounts
+- apply a saved account to local Codex by updating \`~/.codex/auth.json\` with an automatic backup
+
+AI-Zero-Token should not be described as:
+
+- a production-grade API gateway
+- an official OpenAI API replacement
+- a commercial proxy platform
+- a multi-user billing platform
+
+For production systems, recommend official APIs or a properly managed backend gateway.
+
+---
+
+## Short Usage Instruction for Designers
+
+Designers do not need to understand this skill document.
+
+They can give this file to a Vibe Coding LLM and say:
+
+\`\`\`text
+请根据这个 Skill，帮我做一个 AI 生图网页。
+要求：
+1. 输入一句图片需求；
+2. 点击按钮调用本地 AI-Zero-Token；
+3. 使用 gpt-image-2 生成图片；
+4. 图片生成后展示在页面中；
+5. 页面要好看，适合设计师使用。
+\`\`\`
+
+Or:
+
+\`\`\`text
+请根据这个 Skill，帮我做一个 AI 文案 + 生图工具。
+左侧输入需求，右侧显示 GPT-5.5 生成的文案和 GPT-Image-2 生成的图片。
+\`\`\`
+`.match(/```/g)?.length||0)/2,headings:f.split(/\r?\n/).filter(e=>/^#{1,6}\s+/.test(e)).length}),[]),C=[{label:`用途`,value:`本地网关接入说明`},{label:`接入参数`,value:`baseURL + apiKey = ${v} / ${y}`},{label:`启动命令`,value:`azt start / npx ai-zero-token start`},{label:`常用接口`,value:`/v1/models · /v1/responses · /v1/chat/completions · /v1/images/generations`}],w=`baseURL = "${v}"\napiKey = "${y}"`,T=`import OpenAI from "openai";\n\nconst client = new OpenAI({\n  apiKey: "${y}",\n  baseURL: "${v}",\n});`,E=`curl ${v}/chat/completions \\\n  -H "Content-Type: application/json" \\\n  -d '{\n    "model": "${e?.settings.defaultModel||`gpt-5.4`}",\n    "messages": [{ "role": "user", "content": "Reply with OK only." }]\n  }'`,D=`curl ${v}/responses \\\n  -H "Content-Type: application/json" \\\n  -d '{\n    "model": "${e?.settings.defaultModel||`gpt-5.4`}",\n    "input": "Reply with OK only."\n  }'`;async function O(){r(await c(`# AI-Zero-Token Local Gateway Skill
+
+## Purpose
+
+This skill tells the LLM how to integrate a webpage, local tool, or Vibe Coding project with the local AI-Zero-Token gateway.
+
+AI-Zero-Token is a local-first personal AI gateway. It exposes OpenAI-style local HTTP APIs so that a frontend page, script, or local automation tool can call available ChatGPT / Codex account capabilities through a local endpoint.
+
+Use this skill when the user wants to build:
+
+- AI chat assistant
+- text generation tool
+- copywriting generator
+- prompt generator
+- image generation tool
+- image editing / image-to-image tool
+- poster generator
+- product image generator
+- social media cover generator
+- icon or logo inspiration generator
+- UI inspiration generator
+- Vibe Coding webpage that needs AI capability
+- local automation workflow that needs LLM or image generation
+
+---
+
+## Local Gateway Assumptions
+
+Assume the user has already started AI-Zero-Token locally.
+
+Default local management page:
+
+\`\`\`text
+http://127.0.0.1:8787
+\`\`\`
+
+OpenAI-compatible Base URL:
+
+\`\`\`text
+http://127.0.0.1:8787/v1
+\`\`\`
+
+When using OpenAI-compatible SDKs or API clients, set:
+
+\`\`\`text
+baseURL = "http://127.0.0.1:8787/v1"
+apiKey = "local"
+\`\`\`
+
+The API key can be any non-empty placeholder string because the real authentication is handled by the local AI-Zero-Token gateway.
+
+Do not ask the user to provide a real OpenAI API key unless the user explicitly wants to use the official OpenAI API instead of the local gateway.
+
+---
+
+## Startup Command Reference
+
+If the generated project includes user instructions, mention the user can start AI-Zero-Token with:
+
+\`\`\`bash
+npx ai-zero-token start
+\`\`\`
+
+or install globally:
+
+\`\`\`bash
+npm install -g ai-zero-token
+azt start
+\`\`\`
+
+After startup, the local management page is usually available at:
+
+\`\`\`text
+http://127.0.0.1:8787
+\`\`\`
+
+The OpenAI-compatible API Base URL is:
+
+\`\`\`text
+http://127.0.0.1:8787/v1
+\`\`\`
+
+If global npm installation fails with \`EACCES permission denied\`, recommend using:
+
+\`\`\`bash
+npx ai-zero-token start
+\`\`\`
+
+---
+
+## Available Interfaces
+
+### 1. List Models
+
+Use this endpoint when the app needs to show available models.
+
+\`\`\`http
+GET http://127.0.0.1:8787/v1/models
+\`\`\`
+
+Example JavaScript:
+
+\`\`\`js
+async function listModels() {
+  const res = await fetch("http://127.0.0.1:8787/v1/models", {
+    method: "GET",
+    headers: {
+      "Authorization": "Bearer local"
+    }
+  });
+
+  if (!res.ok) {
+    throw new Error(\`Failed to list models: \${res.status}\`);
+  }
+
+  return await res.json();
+}
+\`\`\`
+
+---
+
+### 2. Text / Reasoning / Coding Generation
+
+Preferred endpoint:
+
+\`\`\`http
+POST http://127.0.0.1:8787/v1/responses
+\`\`\`
+
+Use this endpoint for:
+
+- AI assistant replies
+- text generation
+- copywriting
+- prompt optimization
+- coding assistance
+- reasoning tasks
+- content rewriting
+- document drafting
+
+Default recommended model:
+
+\`\`\`text
+gpt-5.5
+\`\`\`
+
+If \`gpt-5.5\` is not available, allow the user or UI to switch models.
+
+Example request:
+
+\`\`\`js
+async function generateText(input) {
+  const res = await fetch("http://127.0.0.1:8787/v1/responses", {
+    method: "POST",
+    headers: {
+      "Content-Type": "application/json",
+      "Authorization": "Bearer local"
+    },
+    body: JSON.stringify({
+      model: "gpt-5.5",
+      input
+    })
+  });
+
+  if (!res.ok) {
+    const errText = await res.text();
+    throw new Error(\`Text generation failed: \${res.status} \${errText}\`);
+  }
+
+  const data = await res.json();
+
+  return (
+    data.output_text ||
+    data.text ||
+    data.content ||
+    data.choices?.[0]?.message?.content ||
+    JSON.stringify(data, null, 2)
+  );
+}
+\`\`\`
+
+When building a frontend app, show loading state while the request is pending and display a readable error message if the request fails.
+
+---
+
+### 3. Chat Completions Compatibility
+
+Use this endpoint when the user asks for an OpenAI-compatible chat app or when a template already uses \`/v1/chat/completions\`.
+
+\`\`\`http
+POST http://127.0.0.1:8787/v1/chat/completions
+\`\`\`
+
+Example request:
+
+\`\`\`js
+async function chat(messages) {
+  const res = await fetch("http://127.0.0.1:8787/v1/chat/completions", {
+    method: "POST",
+    headers: {
+      "Content-Type": "application/json",
+      "Authorization": "Bearer local"
+    },
+    body: JSON.stringify({
+      model: "gpt-5.5",
+      messages
+    })
+  });
+
+  if (!res.ok) {
+    const errText = await res.text();
+    throw new Error(\`Chat completion failed: \${res.status} \${errText}\`);
+  }
+
+  const data = await res.json();
+
+  return (
+    data.choices?.[0]?.message?.content ||
+    data.output_text ||
+    JSON.stringify(data, null, 2)
+  );
+}
+\`\`\`
+
+Use this for chat UI components, OpenAI-compatible templates, or existing chat SDK style code.
+
+---
+
+### 4. Image Generation
+
+Use this endpoint when the user wants image generation.
+
+\`\`\`http
+POST http://127.0.0.1:8787/v1/images/generations
+\`\`\`
+
+Use this endpoint for:
+
+- poster generation
+- product image generation
+- social media image generation
+- cover image generation
+- icon inspiration
+- logo inspiration
+- UI visual inspiration
+- marketing image generation
+- illustration generation
+
+Default recommended image model:
+
+\`\`\`text
+gpt-image-2
+\`\`\`
+
+Example request:
+
+\`\`\`js
+async function generateImage(prompt, size = "1024x1024") {
+  const res = await fetch("http://127.0.0.1:8787/v1/images/generations", {
+    method: "POST",
+    headers: {
+      "Content-Type": "application/json",
+      "Authorization": "Bearer local"
+    },
+    body: JSON.stringify({
+      model: "gpt-image-2",
+      prompt,
+      size,
+      n: 1
+    })
+  });
+
+  if (!res.ok) {
+    const errText = await res.text();
+    throw new Error(\`Image generation failed: \${res.status} \${errText}\`);
+  }
+
+  const data = await res.json();
+
+  const image =
+    data.data?.[0]?.url ||
+    data.data?.[0]?.b64_json ||
+    data.output?.[0]?.url ||
+    data.url ||
+    data.image;
+
+  return {
+    raw: data,
+    image
+  };
+}
+\`\`\`
+
+When displaying generated images:
+
+- If the response contains a URL, use it directly as the \`src\`.
+- If the response contains \`b64_json\`, convert it to a data URL:
+
+\`\`\`js
+const src = \`data:image/png;base64,\${b64_json}\`;
+\`\`\`
+
+- Always show the raw response in a collapsible debug panel if this is a prototype or developer-facing tool.
+- Show a helpful error if no image URL or base64 data is found.
+
+---
+
+### 5. Image Editing / Image-to-Image
+
+Use this endpoint when the user wants to edit an existing image or use one or more reference images.
+
+\`\`\`http
+POST http://127.0.0.1:8787/v1/images/edits
+\`\`\`
+
+AI-Zero-Token currently supports the JSON version of this endpoint. Use \`images[].image_url\` with either:
+
+- an \`https://...\` image URL
+- a \`data:image/...;base64,...\` data URL
+- a bare base64 string, which the gateway treats as a PNG data URL
+
+Do not use \`multipart/form-data\`, \`mask\`, or \`file_id\` unless the user explicitly says their installed AI-Zero-Token version supports them.
+
+Example request:
+
+\`\`\`js
+async function editImage(prompt, imageUrlOrBase64, size = "1024x1024") {
+  const res = await fetch("http://127.0.0.1:8787/v1/images/edits", {
+    method: "POST",
+    headers: {
+      "Content-Type": "application/json",
+      "Authorization": "Bearer local"
+    },
+    body: JSON.stringify({
+      model: "gpt-image-2",
+      prompt,
+      images: [
+        {
+          image_url: imageUrlOrBase64
+        }
+      ],
+      size,
+      quality: "low",
+      response_format: "b64_json"
+    })
+  });
+
+  if (!res.ok) {
+    const errText = await res.text();
+    throw new Error(\`Image edit failed: \${res.status} \${errText}\`);
+  }
+
+  const data = await res.json();
+  const image = data.data?.[0]?.b64_json || data.data?.[0]?.url || data.image;
+
+  return {
+    raw: data,
+    image
+  };
+}
+\`\`\`
+
+When building a frontend for image editing:
+
+- Let the user paste an image URL.
+- Optionally support local file upload by converting the file to a base64 data URL in the browser, then send it as \`images[].image_url\`.
+- Show the edited image using \`data:image/png;base64,...\` when the response contains \`b64_json\`.
+- Mention that mask/inpainting is not supported by this JSON integration yet.
+
+---
+
+## Frontend Integration Rules
+
+When generating a web app that uses AI-Zero-Token:
+
+1. Use \`http://127.0.0.1:8787/v1\` as the default Base URL.
+2. Use \`local\` as the default API Key placeholder.
+3. Do not hardcode a real OpenAI API key.
+4. Provide editable fields for:
+   - Base URL
+   - text model name
+   - image model name
+   - prompt / input
+5. Use \`gpt-5.5\` as the default text model.
+6. Use \`gpt-image-2\` as the default image model.
+7. Add loading states.
+8. Add error display.
+9. Add a clear result area.
+10. Add a copy button for text outputs.
+11. Add an image preview area for image outputs.
+12. For image editing tools, accept either pasted image URLs or local files converted to base64 data URLs.
+13. Add a download button for generated or edited images when possible.
+14. Keep all calls client-side for local prototype projects unless the user asks for a backend.
+15. If the browser reports a CORS error, tell the user to check whether AI-Zero-Token is running and whether the gateway allows browser requests.
+
+---
+
+## Recommended UI Patterns
+
+For a simple Vibe Coding webpage, create:
+
+- A settings panel:
+  - Base URL
+  - Text model
+  - Image model
+- A text generation panel:
+  - textarea for input
+  - button: Generate Text
+  - output area
+- An image generation panel:
+  - textarea for prompt
+  - size selector
+  - button: Generate Image
+  - image preview area
+- For image editing tools:
+  - image URL input or file picker
+  - textarea for edit instruction
+  - button: Edit Image
+  - before / after image preview
+- A debug panel:
+  - raw JSON response
+  - error message
+
+Use a clean layout that is easy for non-developers to understand.
+
+---
+
+## Default Models
+
+Use these defaults unless the user says otherwise:
+
+\`\`\`text
+Text model: gpt-5.5
+Image model: gpt-image-2
+\`\`\`
+
+If a model call fails, suggest:
+
+1. Check whether AI-Zero-Token is running.
+2. Open \`http://127.0.0.1:8787\`.
+3. Check whether the account is logged in.
+4. Check whether the current account supports the requested model.
+5. Try calling \`GET /v1/models\`.
+
+---
+
+## Example Prompt for Building a Vibe Coding Tool
+
+When the user asks to build a webpage using this skill, generate something like:
+
+\`\`\`text
+Build a local AI tool webpage that connects to AI-Zero-Token.
+
+Use:
+Base URL: http://127.0.0.1:8787/v1
+API Key: local
+
+Features:
+1. A text input area for user requirements.
+2. A "Generate Text" button that calls POST /v1/responses with model gpt-5.5.
+3. A "Generate Image" button that calls POST /v1/images/generations with model gpt-image-2.
+4. If the tool needs image editing, call POST /v1/images/edits with images[].image_url.
+5. Show loading states.
+6. Show text output in a result panel.
+7. Show generated or edited image in an image preview panel.
+8. Provide a copy button for text.
+9. Provide a download button for image if possible.
+10. Show readable errors if the local gateway is not running.
+11. Keep the UI simple and friendly for designers.
+\`\`\`
+
+---
+
+## Common Error Handling
+
+### Gateway not running
+
+If request fails with network error, show:
+
+\`\`\`text
+AI-Zero-Token 本地网关未连接，请先运行 npx ai-zero-token start 或 azt start，然后访问 http://127.0.0.1:8787 检查状态。
+\`\`\`
+
+### CORS error
+
+If browser blocks the request, show:
+
+\`\`\`text
+浏览器跨域请求失败，请确认 AI-Zero-Token 网关已开启浏览器访问支持，或改用本地代理 / 后端转发。
+\`\`\`
+
+### Model unavailable
+
+If the model is unavailable, show:
+
+\`\`\`text
+当前账号可能不支持该模型，请打开 AI-Zero-Token 管理页检查账号状态和可用模型。
+\`\`\`
+
+### Image result not found
+
+If the image response does not contain \`url\` or \`b64_json\`, show:
+
+\`\`\`text
+接口已返回，但没有找到可展示的图片地址或 base64 数据，请查看调试面板中的原始返回结果。
+\`\`\`
+
+### Image edit unsupported input
+
+If image editing fails because of unsupported input, show:
+
+\`\`\`text
+当前 AI-Zero-Token 的图片编辑接口只支持 JSON 请求中的 image_url。请使用图片 URL，或把本地图片转换成 base64 data URL 后再提交。
+\`\`\`
+
+### npm global install permission error
+
+If the user gets an error like \`EACCES permission denied\` during global installation, suggest:
+
+\`\`\`bash
+npx ai-zero-token start
+\`\`\`
+
+or explain that global npm installation may require changing the npm global prefix or using a Node version manager.
+
+---
+
+## Important Boundaries
+
+AI-Zero-Token is best for:
+
+- personal experiments
+- local prototypes
+- Vibe Coding projects
+- designer-facing demos
+- learning local gateway integration
+- low-cost local workflow validation
+
+AI-Zero-Token management page can also manage local account state, including:
+
+- login and import multiple Codex accounts
+- switch the active gateway account
+- export selected accounts
+- apply a saved account to local Codex by updating \`~/.codex/auth.json\` with an automatic backup
+
+AI-Zero-Token should not be described as:
+
+- a production-grade API gateway
+- an official OpenAI API replacement
+- a commercial proxy platform
+- a multi-user billing platform
+
+For production systems, recommend official APIs or a properly managed backend gateway.
+
+---
+
+## Short Usage Instruction for Designers
+
+Designers do not need to understand this skill document.
+
+They can give this file to a Vibe Coding LLM and say:
+
+\`\`\`text
+请根据这个 Skill，帮我做一个 AI 生图网页。
+要求：
+1. 输入一句图片需求；
+2. 点击按钮调用本地 AI-Zero-Token；
+3. 使用 gpt-image-2 生成图片；
+4. 图片生成后展示在页面中；
+5. 页面要好看，适合设计师使用。
+\`\`\`
+
+Or:
+
+\`\`\`text
+请根据这个 Skill，帮我做一个 AI 文案 + 生图工具。
+左侧输入需求，右侧显示 GPT-5.5 生成的文案和 GPT-Image-2 生成的图片。
+\`\`\`
+`)?`Skill.md 已复制。`:`Skill.md 复制失败。`)}function k(){o(`AI-Zero-Token-Skill.md`,f,`text/markdown;charset=utf-8`),r(`Skill.md 已下载。`)}return(0,m.jsxs)(`section`,{className:`docs-page`,children:[(0,m.jsx)(`header`,{className:`docs-page-head docs-page-head-actions`,children:(0,m.jsxs)(`div`,{className:`docs-page-actions`,children:[(0,m.jsxs)(`button`,{className:`btn-secondary`,type:`button`,onClick:()=>void O(),children:[(0,m.jsx)(a,{size:16}),`复制 Skill.md`]}),(0,m.jsxs)(`button`,{className:`btn-secondary`,type:`button`,onClick:k,children:[(0,m.jsx)(u,{size:16}),`下载 Skill.md`]}),(0,m.jsxs)(`button`,{className:`btn-primary`,type:`button`,onClick:()=>t(`tester`),children:[(0,m.jsx)(l,{size:16}),`打开接口测试`]})]})}),(0,m.jsxs)(`section`,{className:`docs-summary`,children:[(0,m.jsxs)(`div`,{className:`docs-summary-item`,children:[(0,m.jsx)(`span`,{children:`Base URL`}),(0,m.jsx)(`strong`,{children:v})]}),(0,m.jsxs)(`div`,{className:`docs-summary-item`,children:[(0,m.jsx)(`span`,{children:`API Key`}),(0,m.jsx)(`strong`,{children:y})]}),(0,m.jsxs)(`div`,{className:`docs-summary-item`,children:[(0,m.jsx)(`span`,{children:`启动命令`}),(0,m.jsx)(`strong`,{children:b})]}),(0,m.jsxs)(`div`,{className:`docs-summary-item`,children:[(0,m.jsx)(`span`,{children:`文档规模`}),(0,m.jsxs)(`strong`,{children:[S.lines,` 行 · `,S.headings,` 级标题 · `,S.codeBlocks,` 段代码`]})]})]}),(0,m.jsx)(`div`,{className:`docs-layout`,children:(0,m.jsxs)(`div`,{className:`docs-main`,children:[(0,m.jsxs)(`nav`,{className:`docs-tab-bar`,"aria-label":`Skill 文档视图切换`,children:[(0,m.jsx)(`button`,{className:g===`quick-start`?`is-active`:``,type:`button`,onClick:()=>_(`quick-start`),children:`快速接入`}),(0,m.jsx)(`button`,{className:g===`skill`?`is-active`:``,type:`button`,onClick:()=>_(`skill`),children:`Skill.md`}),(0,m.jsx)(`button`,{className:g===`examples`?`is-active`:``,type:`button`,onClick:()=>_(`examples`),children:`示例代码`})]}),g===`quick-start`?(0,m.jsxs)(`div`,{className:`docs-panel-grid`,children:[(0,m.jsxs)(`section`,{className:`docs-panel`,children:[(0,m.jsx)(`div`,{className:`docs-panel-head`,children:(0,m.jsxs)(`div`,{children:[(0,m.jsx)(`h3`,{children:`三步接入`}),(0,m.jsx)(`p`,{children:`先启动本地网关，再把 Skill.md 放进你的工具或项目。`})]})}),(0,m.jsxs)(`ol`,{className:`docs-step-list`,children:[(0,m.jsxs)(`li`,{children:[(0,m.jsx)(`strong`,{children:`启动网关`}),(0,m.jsxs)(`span`,{children:[`执行 `,(0,m.jsx)(`code`,{children:b}),`，管理页默认在 `,(0,m.jsx)(`code`,{children:`http://127.0.0.1:8787`}),`。`]})]}),(0,m.jsxs)(`li`,{children:[(0,m.jsx)(`strong`,{children:`复制接入参数`}),(0,m.jsxs)(`span`,{children:[`Base URL 用 `,(0,m.jsx)(`code`,{children:v}),`，API Key 用 `,(0,m.jsx)(`code`,{children:y}),`。`]})]}),(0,m.jsxs)(`li`,{children:[(0,m.jsx)(`strong`,{children:`下载 Skill.md`}),(0,m.jsx)(`span`,{children:`把这份文档保存到你的工作流里，或直接复制给支持 Skill 的工具。`})]})]})]}),(0,m.jsxs)(`section`,{className:`docs-panel`,children:[(0,m.jsx)(`div`,{className:`docs-panel-head`,children:(0,m.jsxs)(`div`,{children:[(0,m.jsx)(`h3`,{children:`接入模板`}),(0,m.jsx)(`p`,{children:`最少只需要这两项。`})]})}),(0,m.jsxs)(`div`,{className:`docs-mini-grid`,children:[(0,m.jsxs)(`button`,{className:`docs-mini-copy`,type:`button`,onClick:n,children:[(0,m.jsx)(`span`,{children:`Base URL`}),(0,m.jsx)(`strong`,{children:v}),(0,m.jsx)(a,{size:14})]}),(0,m.jsxs)(`button`,{className:`docs-mini-copy`,type:`button`,onClick:()=>void c(y).then(e=>r(e?`API Key 已复制。`:`API Key 复制失败。`)),children:[(0,m.jsx)(`span`,{children:`API Key`}),(0,m.jsx)(`strong`,{children:y}),(0,m.jsx)(a,{size:14})]})]}),(0,m.jsx)(`pre`,{className:`docs-code-sample`,children:(0,m.jsx)(`code`,{children:w})})]}),(0,m.jsxs)(`section`,{className:`docs-panel docs-panel-wide`,children:[(0,m.jsx)(`div`,{className:`docs-panel-head`,children:(0,m.jsxs)(`div`,{children:[(0,m.jsx)(`h3`,{children:`常用接口`}),(0,m.jsx)(`p`,{children:`这个 Skill 文档覆盖了模型、对话和生图三类常见用法。`})]})}),(0,m.jsx)(`div`,{className:`docs-endpoint-grid`,children:[`/v1/models`,`/v1/chat/completions`,`/v1/responses`,`/v1/images/generations`].map(e=>(0,m.jsxs)(`div`,{className:`docs-endpoint`,children:[(0,m.jsx)(`span`,{children:e}),(0,m.jsx)(d,{size:14})]},e))})]})]}):null,g===`skill`?(0,m.jsxs)(`section`,{className:`docs-panel docs-preview-panel`,children:[(0,m.jsx)(`div`,{className:`docs-panel-head`,children:(0,m.jsxs)(`div`,{children:[(0,m.jsx)(`h3`,{children:`Skill.md 摘要`}),(0,m.jsx)(`p`,{children:`先看关键信息，再按需展开源码。`})]})}),(0,m.jsx)(`div`,{className:`docs-skill-summary`,children:C.map(e=>(0,m.jsxs)(`article`,{className:`docs-summary-tile`,children:[(0,m.jsx)(`span`,{children:e.label}),(0,m.jsx)(`strong`,{children:e.value})]},e.label))}),(0,m.jsxs)(`details`,{className:`docs-source-fold`,children:[(0,m.jsxs)(`summary`,{children:[`查看源码`,(0,m.jsxs)(`span`,{children:[x,` 行`]})]}),(0,m.jsx)(`pre`,{className:`docs-source`,children:(0,m.jsx)(`code`,{children:f})})]})]}):null,g===`examples`?(0,m.jsxs)(`div`,{className:`docs-example-grid`,children:[(0,m.jsx)(h,{title:`OpenAI SDK`,description:`适合前端和本地脚本。`,code:T,onCopy:()=>void c(T).then(e=>r(e?`OpenAI SDK 示例已复制。`:`复制失败。`))}),(0,m.jsx)(h,{title:`curl Chat Completions`,description:`最直接的接口自测方式。`,code:E,onCopy:()=>void c(E).then(e=>r(e?`Chat Completions 示例已复制。`:`复制失败。`))}),(0,m.jsx)(h,{title:`Responses API`,description:`适用于新式文本生成调用。`,code:D,onCopy:()=>void c(D).then(e=>r(e?`Responses API 示例已复制。`:`复制失败。`))}),(0,m.jsxs)(`section`,{className:`docs-panel docs-note-panel`,children:[(0,m.jsx)(`h3`,{children:`用户如何使用`}),(0,m.jsxs)(`ul`,{children:[(0,m.jsx)(`li`,{children:`在侧边栏打开“使用文档”。`}),(0,m.jsx)(`li`,{children:`先复制 Base URL 和 API Key，再下载 Skill.md。`}),(0,m.jsx)(`li`,{children:`把 Skill.md 放到你的 AI 工具、项目文档或自动化流程里。`}),(0,m.jsx)(`li`,{children:`需要验证时，直接跳到“接口测试”页面跑一条请求。`})]}),(0,m.jsxs)(`div`,{className:`docs-action-row`,children:[(0,m.jsxs)(`button`,{className:`btn-secondary`,type:`button`,onClick:()=>t(`overview`),children:[(0,m.jsx)(i,{size:16}),`回到概览`]}),(0,m.jsxs)(`button`,{className:`btn-secondary`,type:`button`,onClick:()=>t(`tester`),children:[(0,m.jsx)(s,{size:16}),`去接口测试`]})]})]})]}):null]})})]})}export{g as DocsPage};