### User Script: Model Selection Example

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/model.md

Demonstrates how to list available models, get the default model ID, and create a new conversation with a selected model. Requires `CAT.agent.model` and `CAT.agent.conversation` grants.

```javascript
// ==UserScript==
// @name        模型选择示例
// @grant       CAT.agent.model
// @grant       CAT.agent.conversation
// ==/UserScript==

const models = await CAT.agent.model.list();
const defaultId = await CAT.agent.model.getDefault();

// 展示给用户选择
const selectedModel = models.find(m => m.supportsVision) || models[0];

const conv = await CAT.agent.conversation.create({
  model: selectedModel.id
});
```

--------------------------------

### SkillScript Implementation Example

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/skill-dev.md

A practical example of a SkillScript for fetching weather data. It demonstrates accessing arguments, user configurations, making API calls, and returning results.

```javascript
// ==SkillScript==
// @name        get_weather
// @description 查询指定城市的天气信息
// @param       city string [required] 城市名称
// @param       days number 预报天数
// @timeout     30
// ==SkillScript==

// 1. 通过 arguments[0] 接收 AI 传入的参数
const { city, days = 3 } = arguments[0];

// 2. CAT_CONFIG 提供用户在管理页面填写的 Skill 配置
const apiKey = CAT_CONFIG.apiKey;
const unit = CAT_CONFIG.unit || "celsius";

// 3. 执行业务逻辑
const url = `https://api.openweathermap.org/data/2.5/forecast?q=${city}&cnt=${days}&units=${unit === "celsius" ? "metric" : "imperial"}&appid=${apiKey}`;
const response = await fetch(url);

if (!response.ok) {
  throw new Error(`API 请求失败: ${response.status}`);
}

const data = await response.json();

// 4. 通过 return 将结果返回给 AI
return {
  city: data.city.name,
  country: data.city.country,
  forecasts: data.list.map(item => ({
    date: item.dt_txt,
    temp: item.main.temp,
    description: item.weather[0].description
  }))
};
```

--------------------------------

### Install Skill via URL

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/skill-install.md

Use this URL directly in your browser or paste it into the ScriptCat management page to install a Skill. Ensure the URL points to a valid SKILL.cat.md file.

```url
https://raw.githubusercontent.com/scriptscat/skills/main/browser-automation/SKILL.cat.md
```

--------------------------------

### Install Skill

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/skill.md

Installs or updates a Skill extension from provided content.

```APIDOC
## install — Install Skill

Installs or updates a Skill extension using its core markdown content and optional script/reference files.

### Method
```javascript
await CAT.agent.skills.install(skillMd: string, scripts?: Array<{ name: string, code: string }>, references?: Array<{ name: string, content: string }>)
```

### Parameters
#### Path Parameters
- **skillMd** (string) - Required - The content of the SKILL.cat.md file.
- **scripts** (Array<{ name: string, code: string }>) - Optional - A list of SkillScript files.
- **references** (Array<{ name: string, content: string }>) - Optional - A list of reference files.

### Example
```javascript
await CAT.agent.skills.install(
  `---
name: my-search
description: Custom search tool
---

When the user needs to search, use the search tool.`,
  [{ name: "search.js", code: skillScriptCode }],
  [{ name: "api-docs.md", content: "# API Docs\n..." }]
);
```

### Returns
`SkillRecord` - The installed or updated Skill's record.
```

--------------------------------

### List Installed Skills

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/skill.md

Retrieve a summary of all installed skills. Each summary includes basic information like name, version, and enabled status.

```javascript
const skills = await CAT.agent.skills.list();
```

--------------------------------

### Install a New Skill

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/skill.md

Install a new skill or update an existing one. Provide the skill's main markdown file, optional scripts, and optional reference files.

```javascript
const record = await CAT.agent.skills.install(skillMd, scripts?, references?);
```

```javascript
const record = await CAT.agent.skills.install(
  `---
name: my-search
description: 自定义搜索工具
---

当用户需要搜索时，使用 search 工具。`,
  [{ name: "search.js", code: skillScriptCode }],
  [{ name: "api-docs.md", content: "# API 文档\n..." }]
);
```

--------------------------------

### Package Multiple Scripts with UserSubscribe

Source: https://context7.com/scriptscat/scriptcat.org/llms.txt

Use `==UserSubscribe==` metadata to bundle multiple scripts into a single subscription file. Subscribed scripts install and update silently. This example shows how to list script URLs and metadata for a suite of scripts.

```javascript
// ==UserSubscribe==
// @name         My Script Suite
// @description  Collection of productivity scripts for example.com
// @version      1.2.0
// @author       Dev
// @connect      api.example.com
// @connect      cdn.example.com
// @scriptUrl    https://example.com/scripts/enhancer.user.js
// @scriptUrl    https://example.com/scripts/dark-mode.user.js
// @scriptUrl    https://example.com/scripts/auto-fill.user.js
// ==/UserSubscribe==

// Install via: open the .user.sub.js URL in a browser with ScriptCat installed.
// - First install: shows confirmation dialog
// - Subsequent updates: silent, no confirmation (unless @connect changes)
// - Scripts removed from @scriptUrl list are auto-uninstalled on next update check
```

--------------------------------

### Example: Auto Form Filling User Script

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/dom.md

A complete example demonstrating how to use various DOM agent functions within a userscript to automate form filling, interaction, and submission.

```javascript
// ==UserScript==
// @name        自动表单填写
// @match       https://example.com/form
// @grant       CAT.agent.dom
// ==/UserScript==

// 等待表单加载
await CAT.agent.dom.waitFor("form#signup", { timeout: 5000 });

// 填写表单
await CAT.agent.dom.fill("input[name=username]", "test_user");
await CAT.agent.dom.fill("input[name=email]", "test@example.com");

// 勾选协议
await CAT.agent.dom.click("input[type=checkbox]#agree");

// 截图保存填写结果
await CAT.agent.dom.screenshot({
  selector: "form#signup",
  saveTo: "screenshots/form-filled.png"
});

// 点击提交
const result = await CAT.agent.dom.click("button[type=submit]", { trusted: true });
if (result.navigated) {
  console.log("表单提交成功，跳转到:", result.url);
}
```

--------------------------------

### GM_download Usage Example (Callback)

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/api.md

Demonstrates how to use GM_download with a callback for the onload event and how to abort the download. This is useful for initiating file downloads with custom configurations.

```javascript
// 回调形式
const dl = GM_download({ url: "https://example.com/file.zip", name: "file.zip", onload: () => alert("完成") });
dl.abort();
```

--------------------------------

### List Installed Skills

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/skill.md

Retrieves a summary of all installed Skill extensions.

```APIDOC
## list — List Installed Skills

Retrieves a summary of all installed Skill extensions.

### Method
```javascript
await CAT.agent.skills.list()
```

### Returns
`SkillSummary[]` - An array of skill summary objects.

**SkillSummary Structure:**
| Field | Type | Description |
|------|------|------|
| `name` | `string` | Skill name |
| `description` | `string` | Skill description |
| `version` | `string` | Version number (semver) |
| `toolNames` | `string[]` | List of included SkillScript tool names |
| `referenceNames` | `string[]` | List of included reference file names |
| `hasConfig` | `boolean` | Whether there are configuration fields declared |
| `enabled` | `boolean` | Whether it is enabled (default `true`) |
| `installUrl` | `string` | Installation source URL (for checking updates) |
| `installtime` | `number` | Installation timestamp |
| `updatetime` | `number` | Update timestamp |
```

--------------------------------

### UserSubscribe Metadata Example

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/subscribe.md

This metadata block is used at the beginning of a subscription script. It defines properties like name, description, version, author, connect permissions, and the URLs of scripts to be subscribed to. Use `UserSubscribe` instead of `UserScript` at the file's start.

```javascript
// ==UserSubscribe==
// @name         xxx
// @description  订阅xxx系列脚本
// @version      0.1.0
// @author       You
// @connect      www.baidu.com
// @scriptUrl    https://script.tampermonkey.net.cn/48.user.js
// @scriptUrl    https://script.tampermonkey.net.cn/49.user.js
// ==/UserSubscribe==
```

--------------------------------

### startMonitor

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/dom.md

Starts monitoring DOM changes and dialog events (alert/confirm/prompt) for a specified tab.

```APIDOC
## startMonitor — 开始监控

```javascript
await CAT.agent.dom.startMonitor(tabId);
```

开始监控指定标签页的 DOM 变化和弹窗（alert/confirm/prompt）。
```

--------------------------------

### Full Example: MCP Tool Management

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/mcp.md

Demonstrates adding an MCP server, testing its connection, listing its tools, and then using a tool in a conversation. MCP tools are automatically registered for AI use.

```javascript
// ==UserScript==
// @name        MCP 工具管理
// @match       *://*/*
// @grant       CAT.agent.mcp
// @grant       CAT.agent.conversation
// ==/UserScript==

// 添加一个 MCP 服务器
const server = await CAT.agent.mcp.addServer({
  name: "Knowledge Base",
  url: "https://my-kb.example.com/mcp",
  apiKey: "my-api-key",
  enabled: true
});

// 测试连接
try {
  await CAT.agent.mcp.testConnection(server.id);
  console.log("连接成功！");
} catch (e) {
  console.error("连接失败:", e);
}

// 查看可用工具
const tools = await CAT.agent.mcp.listTools(server.id);
console.log("可用工具:", tools.map(t => t.name));

// 在对话中使用 — MCP 工具会自动注册，AI 可以直接调用
const conv = await CAT.agent.conversation.create();
const reply = await conv.chat("请使用知识库搜索关于 ScriptCat 的信息");

```

--------------------------------

### Create and Chat with an Agent Conversation

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/index.md

A basic example demonstrating how to create a new conversation with the Agent and send a message. Ensure the `CAT.agent.conversation` permission is granted.

```javascript

// ==UserScript==
// @name        Hello Agent
// @match       *://*/*
// @grant       CAT.agent.conversation
// ==/UserScript==

const conv = await CAT.agent.conversation.create();
const reply = await conv.chat("你好，请介绍一下你自己");
console.log(reply.content);

```

--------------------------------

### SkillScript Parameter Definition Examples

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/skill-dev.md

Illustrates different ways to define parameters for SkillScripts, including required strings, enums, optional numbers, and booleans.

```javascript
// 必填字符串参数
// @param city string [required] 城市名称

// 带枚举的字符串参数
// @param unit string [celsius,fahrenheit] 温度单位

// 可选数字参数
// @param days number 预报天数

// 布尔参数
// @param detailed boolean 是否返回详细信息
```

--------------------------------

### Get Application URL with Ingress

Source: https://github.com/scriptscat/scriptcat.org/blob/main/deploy/helm/templates/NOTES.txt

If Ingress is enabled, this snippet shows how to construct the application URL using configured hosts and paths.

```go-template
{{- if .Values.ingress.enabled }}
{{- range $host := .Values.ingress.hosts }}
  {{- range .paths }}
  http{{ if $.Values.ingress.tls }}s{{ end }}://{{ $host.host }}{{ .path }}
  {{- end }}
{{- end }}
{{- end }}
```

--------------------------------

### GM_cookie Usage Example (Promise)

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/api.md

Demonstrates the Promise-based usage of GM_cookie for listing, setting, and deleting cookies using async/await syntax. Requires '@connect' declaration in metadata for the specified domain.

```javascript
// Promise 形式
const cookies = await GM.cookie.list({ url: "https://example.com" });
await GM.cookie.set({ name: "foo", value: "bar", domain: "example.com" });
await GM.cookie.delete("foo", { domain: "example.com" });
```

--------------------------------

### Start DOM Monitoring with startMonitor

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/dom.md

Initiates monitoring for DOM changes and dialog events (alert, confirm, prompt) on a specified tab.

```javascript
await CAT.agent.dom.startMonitor(tabId);
```

--------------------------------

### Get Model List Summary

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/model.md

Obtain a text summary of all configured models, suitable for direct injection into prompts for AI reference.

```javascript
const summary = await CAT.agent.model.getSummary();
```

--------------------------------

### Get MCP Server Prompt

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/mcp.md

Fetches a prompt template from an MCP server, optionally with arguments to fill in parameters.

```javascript
const messages = await CAT.agent.mcp.getPrompt(serverId, promptName, args?);
```

--------------------------------

### GM_cookie Usage Example (Callback)

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/api.md

Demonstrates the callback-based usage of GM_cookie for listing, setting, and deleting cookies. Requires '@connect' declaration in metadata for the specified domain.

```javascript
// 回调形式
GM_cookie("list", { url: "https://example.com" }, (cookies) => {
  console.log(cookies);
  GM_cookie("set", {
    name: "foo",
    value: "bar",
    domain: "example.com"
  }, (result) => {
    console.log(result);
    GM_cookie("delete", { name: "foo", domain: "example.com" }, (result) => {
      console.log(result);
    });
  });
});
```

--------------------------------

### Save Screenshot to OPFS - JavaScript

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/dom.md

Example demonstrating how to save a screenshot directly to the Origin Private File System (OPFS). It logs the save path and file size upon completion.

```javascript
// 保存截图到 OPFS
const shot = await CAT.agent.dom.screenshot({
  saveTo: "screenshots/page.png",
  quality: 90
});
console.log(`已保存到 ${shot.path}，大小 ${shot.size} 字节`);
```

--------------------------------

### Script Subscription Mode (`==UserSubscribe==`)

Source: https://context7.com/scriptscat/scriptcat.org/llms.txt

Package multiple scripts into a single subscription file. Subscribed scripts install and update silently without per-script confirmations.

```APIDOC
## Script Subscription Mode (`==UserSubscribe==`)

Package multiple scripts into a single subscription file. Subscribed scripts install and update silently without per-script confirmations.

```js
// ==UserSubscribe==
// @name         My Script Suite
// @description  Collection of productivity scripts for example.com
// @version      1.2.0
// @author       Dev
// @connect      api.example.com
// @connect      cdn.example.com
// @scriptUrl    https://example.com/scripts/enhancer.user.js
// @scriptUrl    https://example.com/scripts/dark-mode.user.js
// @scriptUrl    https://example.com/scripts/auto-fill.user.js
// ==/UserSubscribe==

// Install via: open the .user.sub.js URL in a browser with ScriptCat installed.
// - First install: shows confirmation dialog
// - Subsequent updates: silent, no confirmation (unless @connect changes)
// - Scripts removed from @scriptUrl list are auto-uninstalled on next update check
```
```

--------------------------------

### SkillScript Metadata Example

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/skill-dev.md

Define SkillScript metadata using special comments. This includes the tool's name, description, parameters, required permissions, external libraries, and timeout.

```javascript
// ==SkillScript==
// @name        get_weather
// @description 查询指定城市的天气信息
// @param       city string [required] 城市名称，支持中英文
// @param       days number 预报天数，默认3天
// @param       format string [json,text] 输出格式
// @grant       CAT.agent.opfs
// @require     https://cdn.example.com/utils.js
// @timeout     60
// ==SkillScript==
```

--------------------------------

### Define User Configuration UI with YAML

Source: https://context7.com/scriptscat/scriptcat.org/llms.txt

Use a YAML block after `==UserConfig==` to define a configuration UI for end users. Values are read using `GM_getValue("group.key")`. This example shows various input types like text, select, number, checkbox, multi-select, and textarea.

```javascript
// ==UserScript==
// @name    Config Demo
// @match   *://*/*
// @grant   GM_getValue
// @grant   CAT_userConfig
// ==/UserScript==

/* ==UserConfig==
general:
  apiKey:
    title: API Key
    description: Your API key for the service
    type: text
    password: true
  language:
    title: Language
    type: select
    default: en
    values: [en, zh, ja, fr, de]
  maxResults:
    title: Max Results
    type: number
    default: 10
    min: 1
    max: 100
    unit: items
  autoRun:
    title: Auto Run on Load
    type: checkbox
    default: false
  tags:
    title: Filter Tags
    type: mult-select
    default: [news, tech]
    values: [news, tech, sports, science, entertainment]
  notes:
    title: Custom Notes
    type: textarea
    default: ""
==/UserConfig== */

// Read configuration values using "group.key" path
const apiKey   = GM_getValue("general.apiKey", "");
const lang     = GM_getValue("general.language", "en");
const maxItems = GM_getValue("general.maxResults", 10);
const autoRun  = GM_getValue("general.autoRun", false);
const tags     = GM_getValue("general.tags", []);

console.log({ apiKey, lang, maxItems, autoRun, tags });

// Open the config UI programmatically
GM_registerMenuCommand("⚙️ Settings", CAT_userConfig);
```

--------------------------------

### getSummary — 获取模型列表摘要

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/model.md

Returns a text summary of all configured models, suitable for direct injection into prompts for AI reference.

```APIDOC
## getSummary — 获取模型列表摘要

### Description
Returns a text summary of all configured models, suitable for direct injection into prompts for AI reference.

### Method
`CAT.agent.model.getSummary()`

### Parameters
None

### Response
#### Success Response (string)
- Returns a string containing a summary of all configured models.

### Response Example
```json
"Available models: GPT-4o (OpenAI), Claude 3 Sonnet (Anthropic)"
```
```

--------------------------------

### Get Skill Details

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/skill.md

Retrieves detailed information about a specific Skill extension.

```APIDOC
## get — Get Skill Details

Retrieves detailed information about a specific Skill extension by its name.

### Method
```javascript
await CAT.agent.skills.get(name: string)
```

### Parameters
#### Path Parameters
- **name** (string) - Required - The name of the Skill to retrieve.

### Returns
`SkillRecord | null` - The complete SkillRecord if found, otherwise `null`.

**SkillRecord Structure:**
Inherits all fields from `SkillSummary` and includes:
| Field | Type | Description |
|------|------|------|
| `prompt` | `string` | The Markdown body content from SKILL.cat.md (prompt for AI) |
| `config` | `Record<string, SkillConfigField>` | Configuration field definitions (schema) |

**SkillConfigField Structure:**
| Field | Type | Description |
|------|------|------|
| `title` | `string` | Display title |
| `type` | `"text" | "number" | "select" | "switch"` | Field type |
| `secret` | `boolean` | Whether it is sensitive information (masked in UI) |
| `required` | `boolean` | Whether it is required |
| `default` | `unknown` | Default value |
| `values` | `string[]` | List of options (only for `select` type) |
```

--------------------------------

### SKILL.cat.md Metadata and Prompt Format

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/skill-dev.md

Define Skill metadata using YAML frontmatter and provide AI instructions in the Markdown body. The frontmatter includes name, description, and configuration.

```markdown
---
name: "weather-assistant"
description: "天气查询助手，支持全球城市天气查询和预报"
config:
  apiKey:
    title: "OpenWeather API Key"
    type: "text"
    secret: true
    required: true
  unit:
    title: "温度单位"
    type: "select"
    values: ["celsius", "fahrenheit"]
    default: "celsius"
  detailed:
    title: "详细模式"
    type: "switch"
    default: false
  maxDays:
    title: "预报天数"
    type: "number"
    default: 7
---

# 天气查询助手

你可以使用以下工具查询天气信息：

## 工具说明

- **get_weather**: 查询指定城市的当前天气和未来预报
  - 参数 `city` 为城市名称（支持中英文）
  - 参数 `days` 为预报天数

## 使用规则

1. 用户询问天气时，先确认城市名称
2. 默认返回当前天气 + 3 天预报
3. 温度根据配置的单位显示
```

--------------------------------

### addServer

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/mcp.md

Adds a new MCP server configuration.

```APIDOC
## addServer — Add Server

### Description
Adds a new MCP server configuration.

### Method
`CAT.agent.mcp.addServer(config: MCPServerConfig)`

### Parameters
#### Request Body
- **config** (`MCPServerConfig`) - Required - The configuration for the new MCP server.
  - **name** (`string`) - Required - Display name for the server.
  - **url** (`string`) - Required - HTTP endpoint URL of the MCP server.
  - **apiKey** (`string`) - Optional - API Key for authentication.
  - **headers** (`Record<string, string>`) - Optional - Custom request headers.
  - **enabled** (`boolean`) - Optional - Whether the server should be enabled by default (defaults to true).

### Request Example
```json
{
  "name": "My MCP Server",
  "url": "https://mcp.example.com/api",
  "apiKey": "sk-xxx",
  "headers": {
    "X-Custom": "value"
  },
  "enabled": true
}
```
```

--------------------------------

### get — 获取指定模型

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/model.md

Retrieves the details of a specific model by its ID. Returns null if the model is not found.

```APIDOC
## get — 获取指定模型

### Description
Retrieves the details of a specific model by its ID. Returns null if the model is not found.

### Method
`CAT.agent.model.get(modelId)`

### Parameters
#### Path Parameters
- **modelId** (string) - Required - The ID of the model to retrieve.

### Response
#### Success Response (ModelSummary | null)
- **id** (string) - 模型 ID
- **name** (string) - 显示名称
- **provider** ("openai" | "anthropic" | "zhipu") - Provider 类型
- **apiBaseUrl** (string) - API 基础地址
- **model** (string) - 模型标识符（如 `gpt-4o`、`claude-sonnet-4-20250514`）
- **maxTokens** (number) - 最大输出 Token 数
- **contextWindow** (number) - 上下文窗口大小
- **supportsVision** (boolean) - 是否支持视觉输入（图片）
- **supportsImageOutput** (boolean) - 是否支持图片生成

### Response Example
```json
{
  "id": "model-1",
  "name": "GPT-4o",
  "provider": "openai",
  "apiBaseUrl": "https://api.openai.com",
  "model": "gpt-4o",
  "maxTokens": 128000,
  "contextWindow": 128000,
  "supportsVision": true,
  "supportsImageOutput": false
}
```
```

--------------------------------

### Get Default Model ID

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/model.md

Retrieve the ID of the user's currently set default model.

```javascript
const defaultId = await CAT.agent.model.getDefault();
```

--------------------------------

### Skill Directory Structure

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/skill-dev.md

Organize your Skill's components within a dedicated directory. SKILL.cat.md is the mandatory entry point.

```tree
my-skill/
├── SKILL.cat.md          # 必须：元数据 + 提示词（入口文件）
├── scripts/              # 可选：SkillScript 工具脚本
│   ├── search.js
│   └── export.js
└── references/           # 可选：参考资料文件
    ├── api-docs.md
    └── examples.json
```

--------------------------------

### Get MCP Server Details

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/mcp.md

Fetches the configuration details for a specific MCP server using its ID.

```javascript
const server = await CAT.agent.mcp.getServer(serverId);
```

--------------------------------

### Get Application URL with NodePort

Source: https://github.com/scriptscat/scriptcat.org/blob/main/deploy/helm/templates/NOTES.txt

When the service type is NodePort, this command exports the NodePort and Node IP to access the application.

```bash
export NODE_PORT=$(kubectl get --namespace {{ .Release.Namespace }} -o jsonpath="{.spec.ports[0].nodePort}" services {{ include "scriptcat-docs.fullname" . }})
export NODE_IP=$(kubectl get nodes --namespace {{ .Release.Namespace }} -o jsonpath="{.items[0].status.addresses[0].address}")
echo http://$NODE_IP:$NODE_PORT
```

--------------------------------

### CAT_scriptLoaded - Early-Start Completion

Source: https://context7.com/scriptscat/scriptcat.org/llms.txt

When a script uses `@early-start` to inject before page load, use `CAT_scriptLoaded()` to wait until the DOM and dependencies are fully available.

```APIDOC
## `CAT_scriptLoaded` — Early-Start Completion

When a script uses `@early-start` to inject before page load, use `CAT_scriptLoaded()` to wait until the DOM and dependencies are fully available.

```js
// ==UserScript==
// @name         Early Start Demo
// @match        *://*/*
// @run-at       document-start
// @early-start
// @grant        GM_setValue
// ==/UserScript==

// This runs before the page — intercept XHR immediately
const origOpen = XMLHttpRequest.prototype.open;
XMLHttpRequest.prototype.open = function(method, url, ...args) {
  console.log("[early] Intercepted:", method, url);
  return origOpen.call(this, method, url, ...args);
};

// Wait for full script environment before reading storage
CAT_scriptLoaded().then(() => {
  const savedData = GM_getValue("cache", null);
  console.log("Script fully loaded, cache:", savedData);
});
```
```

--------------------------------

### Get Specific Model Details

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/model.md

Retrieve detailed information for a specific model using its ID. Returns `null` if the model does not exist.

```javascript
const model = await CAT.agent.model.get(modelId);
```

--------------------------------

### listServers

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/mcp.md

Lists all configured MCP servers.

```APIDOC
## listServers — List Servers

### Description
Lists all configured MCP servers.

### Method
`CAT.agent.mcp.listServers()`

### Returns
`MCPServerConfig[]` - An array of MCP server configurations.

#### MCPServerConfig Fields
- **id** (`string`) - Server ID
- **name** (`string`) - Display name
- **url** (`string`) - HTTP endpoint URL
- **apiKey** (`string`) - API Key (optional)
- **headers** (`Record<string, string>`) - Custom request headers (optional)
- **enabled** (`boolean`) - Whether the server is enabled
- **createtime** (`number`) - Creation timestamp
- **updatetime** (`number`) - Update timestamp
```

--------------------------------

### list — 列出所有模型

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/model.md

Lists all configured models and their details. The returned list does not include the API key.

```APIDOC
## list — 列出所有模型

### Description
Lists all configured models and their details. The returned list does not include the API key.

### Method
`CAT.agent.model.list()`

### Parameters
None

### Response
#### Success Response (ModelSummary[])
- **id** (string) - 模型 ID
- **name** (string) - 显示名称
- **provider** ("openai" | "anthropic" | "zhipu") - Provider 类型
- **apiBaseUrl** (string) - API 基础地址
- **model** (string) - 模型标识符（如 `gpt-4o`、`claude-sonnet-4-20250514`）
- **maxTokens** (number) - 最大输出 Token 数
- **contextWindow** (number) - 上下文窗口大小
- **supportsVision** (boolean) - 是否支持视觉输入（图片）
- **supportsImageOutput** (boolean) - 是否支持图片生成

### Response Example
```json
[
  {
    "id": "model-1",
    "name": "GPT-4o",
    "provider": "openai",
    "apiBaseUrl": "https://api.openai.com",
    "model": "gpt-4o",
    "maxTokens": 128000,
    "contextWindow": 128000,
    "supportsVision": true,
    "supportsImageOutput": false
  }
]
```
```

--------------------------------

### List All Models

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/model.md

Retrieve a list of all configured models. The returned `ModelSummary[]` does not include the `apiKey` field.

```javascript
const models = await CAT.agent.model.list();
```

--------------------------------

### Background Script Example

Source: https://context7.com/scriptscat/scriptcat.org/llms.txt

Background scripts run persistently in the service worker context without DOM access. They must return a Promise to indicate completion.

```javascript
// ==UserScript==
// @name         Session Heartbeat
// @namespace    https://example.com/
// @version      1.0.0
// @author       Dev
// @background
// @grant        GM_xmlhttpRequest
// @grant        GM_notification
// @connect      api.example.com
// ==/UserScript==

return new Promise((resolve, reject) => {
  GM_xmlhttpRequest({
    method: "POST",
    url: "https://api.example.com/heartbeat",
    headers: { "Content-Type": "application/json" },
    data: JSON.stringify({ ts: Date.now() }),
    onload(res) {
      if (res.status === 200) {
        resolve("ok");
      } else {
        reject(new CATRetryError("Heartbeat failed", 30)); // retry in 30s
      }
    },
    onerror() {
      reject(new CATRetryError("Network error", 60));
    }
  });
});
```

--------------------------------

### Wait for Full Script Environment with CAT_scriptLoaded

Source: https://context7.com/scriptscat/scriptcat.org/llms.txt

When using `@early-start`, `CAT_scriptLoaded()` ensures that the DOM and dependencies are fully available before proceeding. This is useful for scripts that need to interact with the page's environment after initial injection.

```javascript
// ==UserScript==
// @name         Early Start Demo
// @match        *://*/*
// @run-at       document-start
// @early-start
// @grant        GM_setValue
// ==/UserScript==

// This runs before the page — intercept XHR immediately
const origOpen = XMLHttpRequest.prototype.open;
XMLHttpRequest.prototype.open = function(method, url, ...args) {
  console.log("[early] Intercepted:", method, url);
  return origOpen.call(this, method, url, ...args);
};

// Wait for full script environment before reading storage
CAT_scriptLoaded().then(() => {
  const savedData = GM_getValue("cache", null);
  console.log("Script fully loaded, cache:", savedData);
});
```

--------------------------------

### listPrompts

Source: https://github.com/scriptscat/scriptcat.org/blob/main/docs/dev/agent/mcp.md

Lists the prompt templates available on a specific MCP server.

```APIDOC
## listPrompts — List Prompts

### Description
Lists the prompt templates available on a specific MCP server.

### Method
`CAT.agent.mcp.listPrompts(serverId: string)`

### Parameters
#### Path Parameters
- **serverId** (`string`) - Required - The ID of the MCP server.

### Returns
`Prompt[]` - An array of prompt template objects.
```