doc: en and cn with examples (#24)

* doc: cn example

* doc: packages

* doc: en and cn

* doc: more details

docs/pages/en/guide/getting-started.md

@@ -1 +1,154 @@
 # Getting Started
+
+## Installation
+
+Install Neuri using your favorite package manager:
+
+```bash
+pnpm add neuri
+```
+
+## Basic Usage
+
+```ts
+import { env } from 'node:process'
+import {
+  composeAgent,
+  defineToolFunction,
+  resolveFirstTextContentFromChatCmpl,
+  system,
+  toolFunction,
+  user,
+} from 'neuri/openai'
+import OpenAI from 'openai'
+import * as z from 'zod'
+
+// Create OpenAI client instance
+const openai: OpenAI = new OpenAI({
+  baseURL: env.OPENAI_API_BASEURL,
+  apiKey: env.OPENAI_API_KEY,
+})
+
+// Compose Agent and tool functions
+const { call } = composeAgent({
+  openAI: openai,
+  tools: [
+    // Define a tool function to get city
+    defineToolFunction<Record<string, never>, string>(
+      await toolFunction('getCity', 'Get the user\'s city', z.object({})),
+      async () => {
+        return 'New York City'
+      }
+    ),
+    // Define a tool function to get weather
+    defineToolFunction<{ cityCode: string }, { city: string, weather: string, degreesCelsius: number }>(
+      await toolFunction('getWeather', 'Get the current weather', z.object({
+        cityCode: z.string().min(1).describe('City code to get weather for')
+      })),
+      async ({ parameters: { cityCode } }) => {
+        return {
+          city: 'New York City',
+          weather: 'sunny',
+          degreesCelsius: 26
+        }
+      }
+    )
+  ]
+})
+
+// Call Agent for conversation
+const response = await call([
+  system('I am a helpful assistant that can provide weather information.'),
+  user('What is the weather like today?')
+], {
+  model: 'openai/gpt-3.5-turbo'
+})
+
+const result = resolveFirstTextContentFromChatCmpl(response)
+console.log(result)
+```
+
+## Extension Packages
+
+### @neuri/use-fs
+
+Filesystem related tool functions package, supporting local filesystem and GitHub repository operations.
+
+```ts
+import { FileSystem } from '@neuri/use-fs/node'
+import { GitHubPublicFileSystem } from '@neuri/use-fs/github'
+
+// Local filesystem
+const fs = await FileSystem()
+const { readFile, writeFile, listFilesInDirectory } = fs
+
+// GitHub repository operations
+const github = await GitHubPublicFileSystem()
+const { listFilesInDirectory: listGitHubFiles, readFile: readGitHubFile } = github
+```
+
+### @neuri/use-search
+
+Search engine related tool functions package, currently supporting Google search (via SerpAPI).
+
+```ts
+import { SerpApi } from '@neuri/use-search/serpapi'
+
+const serpapi = await SerpApi({
+  apiKey: 'your-serpapi-key'
+})
+
+const { searchGoogle } = serpapi
+```
+
+### @neuri/format-code
+
+Code formatting and syntax highlighting related tool functions package.
+
+```ts
+import { tokenizeByTextMateGrammar } from '@neuri/format-code/textmate'
+
+// Use TextMate grammar to parse code
+const result = await tokenizeByTextMateGrammar('typescript', code)
+```
+
+## API Reference
+
+### Core APIs
+
+#### neuri()
+
+Create a new Neuri builder instance.
+
+#### agent(name?: string)
+
+Create a new Agent builder instance.
+
+Parameters:
+- `name`: Agent name, optional. Defaults to `__default__`
+
+#### composeAgent(options)
+
+Compose Agent and tool functions.
+
+Parameters:
+- `options.openAI`: OpenAI client instance
+- `options.tools`: Array of tool functions
+
+#### defineToolFunction(tool, func, options?)
+
+Define a tool function.
+
+Parameters:
+- `tool`: Tool function definition created by `toolFunction()`
+- `func`: Tool function implementation
+- `options`: Optional configuration including provider and hooks
+
+#### toolFunction(name, description, parameters)
+
+Create a tool function definition.
+
+Parameters:
+- `name`: Tool function name
+- `description`: Tool function description
+- `parameters`: Parameter Schema, supports zod or valibot

docs/pages/en/guide/index.md

@@ -1 +1,77 @@
 # Overview
+
+Neuri is a simple and easy-to-use AI Agent framework that provides a complete toolkit to help you quickly build OpenAI-based AI applications.
+
+## Features
+
+- 📦 **100% TypeScript Support** - Complete type definitions out of the box
+- ✅ **High Test Coverage** - Over 70% test coverage ensures reliability
+- 🧠 **Model Agnostic** - Structured data support for various models including Llama 3.1, Phi3.5 Mistral, OpenAI, etc.
+- 📃 **JSON Structured Data**
+  - Works with streaming responses
+  - Clean and type-safe data handling
+- 💻 **Code Snippet Extraction**
+  - Supports streaming extraction
+  - Path-based code snippet handling
+  - Powered by VSCode TextMate grammar
+- 👷 **Easy Agent Composition**
+  - Simple API for declaring and composing agents
+  - Flexible function composition
+  - Built-in tool function system
+- 📚 **Rich Component Library**
+  - FileSystem operations (local & GitHub)
+  - Search engine integration
+  - Code formatting utilities
+
+### Simple Agent Framework
+
+Neuri is built on OpenAI's Function Calling feature and provides a simple agent framework. You just need to define tool functions, and then let AI automatically choose the right tools to complete tasks.
+
+```ts
+const agent = await neuri()
+  .agent(
+    agent('weather')
+      .tool('getCurrentLocation', object({}), async () => 'Shanghai')
+      .tool('getCurrentWeather', object({ location: string() }),
+        async ({ parameters: { location } }) => {
+          return `${location}, China: 22 degree Celsius`
+        })
+      .build()
+  )
+  .build({
+    provider: {
+      apiKey: process.env.OPENAI_API_KEY!,
+      baseURL: process.env.OPENAI_API_BASEURL!,
+    },
+  })
+```
+
+### Rich Tool Functions
+
+Neuri provides multiple extension packages, including:
+
+- `@neuri/use-fs`: Filesystem operations, supporting local filesystem and GitHub repositories
+- `@neuri/use-search`: Search engine integration, supporting Google search
+- `@neuri/format-code`: Code formatting and syntax highlighting
+
+### Type Safe
+
+Neuri is written in TypeScript and provides complete type definitions. Tool function parameters and return values are type-checked to make your code more robust.
+
+### Modular Design
+
+Neuri uses a modular design where core functionality and extension packages are separated. You can import only the features you need to reduce bundle size.
+
+## Related Projects
+
+- [neuri-go](https://github.com/lingticio/neuri-go) - Go implementation of Neuri
+- [llmg](https://github.com/lingticio/llmg) - Powerful LLM gateway for building LLM applications
+- [devtools](https://github.com/guiiai/devtools) - Frontend DevTools with LLM copilot features
+- [ollama-operator](https://github.com/nekomeowww/ollama-operator) - Kubernetes operator for Ollama
+- [nolebase/integrations](https://github.com/nolebase/integrations) - Documentation integrations powered by VitePress
+
+## Next Steps
+
+- [Getting Started](/pages/en/guide/getting-started): Learn how to install and use Neuri
+- [GitHub](https://github.com/lingticio/neuri-js): Visit the project repository
+- [Discord](https://discord.gg/link-to-your-discord): Join the community discussion

docs/pages/en/index.md

@@ -4,6 +4,8 @@ layout: home
 
 hero:
   name: Neuri
+  text: Simple and Easy AI Agent Framework
+  tagline: A unified solution for structured data manipulation, agent composition, code editing, filesystem and more
   actions:
     - theme: brand
       text: Get Started
@@ -14,4 +16,24 @@ hero:
     - theme: alt
       text: View on GitHub
       link: https://github.com/lingticio/neuri-js
+
+features:
+  - icon: 🤖
+    title: 100% TypeScript Support
+    details: Complete type definitions out of the box, with over 70% test coverage
+  - icon: 🧠
+    title: Model Agnostic
+    details: Support for various models including Llama, Phi, Mistral, OpenAI, with structured data handling
+  - icon: 💻
+    title: Code Snippet Extraction
+    details: Path-based code snippet handling with streaming support, powered by VSCode TextMate grammar
+  - icon: 👷
+    title: Easy Agent Composition
+    details: Simple API for declaring and composing agents with flexible function composition
+  - icon: 📚
+    title: Rich Component Library
+    details: FileSystem operations, search engine integration, code formatting and more
+  - icon: 🚀
+    title: Quick to Start
+    details: Get started with just a few lines of code, with comprehensive documentation and examples
 ---