### Ollama Setup Commands

Source: https://github.com/anrgct/autodev-codebase/blob/master/CONFIG.md

Commands to install Ollama, pull models, configure Autodev for Ollama, and start indexing.

```bash
# Install and start Ollama
brew install ollama
ollama serve

# Pull embedding model
ollama pull nomic-embed-text

# Configure
codebase config --set embedderProvider=ollama,embedderModelId=nomic-embed-text
codebase config --set qdrantUrl=http://localhost:6333

# Start indexing
codebase index --log-level=debug --force
```

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

### Quick Start: MCP Server Demo

Source: https://github.com/anrgct/autodev-codebase/blob/master/README.md

Demonstrates running the MCP server in demo mode. This command starts the server with demo data for testing.

```bash
# MCP server
codebase index --serve --demo
```

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

### Basic Node.js Usage Example

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/project-outline.md

Demonstrates fundamental usage patterns for the autodev library in a Node.js environment. No specific setup is required beyond having the library installed.

```typescript
function basicUsageExample() {
  // ... implementation details ...
}
```

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

### Project Configuration Example

Source: https://github.com/anrgct/autodev-codebase/blob/master/CONFIG.md

Example of a project-specific configuration file (`./autodev-config.json`). This file allows you to define settings for a particular project.

```json
{
  "embedderProvider": "ollama",
  "embedderModelId": "nomic-embed-text",
  "qdrantUrl": "http://localhost:6333"
}
```

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

### Global Configuration Example

Source: https://github.com/anrgct/autodev-codebase/blob/master/CONFIG.md

Example of a global user-specific configuration file (`~/.autodev-cache/autodev-config.json`). These settings apply across all projects.

```json
{
  "embedderProvider": "openai",
  "embedderModelId": "text-embedding-3-small",
  "embedderOpenAiApiKey": "sk-your-key",
  "qdrantUrl": "http://localhost:6333",
  "vectorSearchMinScore": 0.3,
  "vectorSearchMaxResults": 20
}
```

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

### Call Command Framework Setup

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/plans/260117-dependency-cli.md

This snippet shows the initial setup for the 'call' command, including file creation, function implementation, and command registration within the CLI.

```typescript
// src/commands/call.ts
// 实现 createCallCommand() 函数
// 在 src/cli.ts 中注册新命令
```

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

### Install CLI and Configure Embedding Backend

Source: https://context7.com/anrgct/autodev-codebase/llms.txt

Install the CLI globally and configure an embedding backend. For privacy, Ollama is recommended for local operation. Ensure Qdrant is running.

```bash
# Option A: Fully local with Ollama (recommended for privacy)
brew install ollama ripgrep
ollama serve
ollama pull nomic-embed-text

# Start Qdrant vector database
docker run -d -p 6333:6333 -p 6334:6334 --name qdrant qdrant/qdrant

# Install CLI
npm install -g @autodev/codebase

# Configure for Ollama
codebase config --set embedderProvider=ollama,embedderModelId=nomic-embed-text
codebase config --set qdrantUrl=http://localhost:6333

# Option B: OpenAI
export OPENAI_API_KEY="sk-your-key"
codebase config --set embedderProvider=openai,embedderModelId=text-embedding-3-small
codebase config --set qdrantUrl=http://localhost:6333
```

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

### User Guide: Dependency Analysis Capabilities

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/plans/260121-top-level-calls-not-tracked.md

This user guide update outlines the supported dependency types, including the new 'module-level calls', and provides an example of how top-level code in files like `app.js` is now tracked.

```markdown
## 依赖分析功能

### 支持的依赖类型

1. **函数间调用** - `functionA → functionB`
2. **类间调用** - `ClassA → ClassB`
3. **方法间调用** - `ClassA.methodX → ClassB.methodY`
4. **模块级调用** - `module → function/class` (新增)

### 顶层代码支持

依赖分析器会自动追踪文件顶层代码的函数调用：

```javascript
// 入口文件 app.js
const service = new UserService();  // ✅ 会追踪
service.initialize();               // ✅ 会追踪
console.log('Ready');               // ❌ 自动过滤（内置函数）
```

这对以下场景特别有用：
- 入口文件的初始化逻辑
- 脚本文件的执行流程
- 测试文件的顶层 describe/it 调用
```

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

### CLI Example for Autodev

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/project-outline.md

Demonstrates how to interact with the autodev tools via the command-line interface. This example is useful for understanding command-line arguments and output.

```typescript
function cliExample() {
  // ... implementation details ...
}
```

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

### Start Qdrant Vector Database

Source: https://github.com/anrgct/autodev-codebase/blob/master/README.md

Run a Qdrant instance using Docker for vector storage. This command starts a container named 'qdrant' and maps the default ports.

```bash
docker run -d -p 6333:6333 -p 6334:6334 --name qdrant qdrant/qdrant
```

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

### Summarizer Language Examples

Source: https://github.com/anrgct/autodev-codebase/blob/master/CONFIG.md

Examples of configuring the summarizer for English and Chinese output.

```json
{
  "summarizerProvider": "ollama",
  "summarizerLanguage": "English"
}
```

```json
{
  "summarizerProvider": "ollama",
  "summarizerLanguage": "Chinese"
}
```

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

### Start Stdio MCP Adapter

Source: https://context7.com/anrgct/autodev-codebase/llms.txt

Starts the stdio adapter, connecting it to the running HTTP MCP server via its URL. This allows IDEs using stdio transport to communicate with the HTTP server.

```bash
codebase stdio --server-url=http://localhost:3001/mcp
```

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

### Test Development Mode

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/plans/260122-unify-wasm-path-resolution.md

Start the development server and manually verify that WASM files are loaded correctly in development mode.

```bash
npm run dev
```

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

### MCP Server Status Output

Source: https://context7.com/anrgct/autodev-codebase/llms.txt

Example output when starting the MCP server, showing the running address, workspace path, MCP endpoint, and health check URL.

```text
🔍 Codebase MCP Server running at http://localhost:3001
📁 Workspace: /my/project
🌐 MCP endpoint: http://localhost:3001/mcp
❤️  Health check: http://localhost:3001/health
```

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

### Start Stdio MCP Adapter with Custom Timeout

Source: https://context7.com/anrgct/autodev-codebase/llms.txt

Starts the stdio adapter with a custom timeout value (in milliseconds) for communication with the HTTP MCP server.

```bash
codebase stdio --server-url=http://localhost:3001/mcp --timeout=60000
```

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

### Quick Start: Index and Search Demo

Source: https://github.com/anrgct/autodev-codebase/blob/master/README.md

Demonstrates the core indexing and search functionality in demo mode. This is recommended for first-time users to test the tool without a real project.

```bash
# Index & search
codebase index --demo
codebase search "user greet" --demo
```

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

### Autodev Codebase Configuration Examples (JSON)

Source: https://github.com/anrgct/autodev-codebase/blob/master/README.md

Illustrative JSON configurations for different embedder providers (Ollama, OpenAI, OpenAI-compatible). These examples show how to set provider, model ID, API keys, and base URLs.

```json
{
  "embedderProvider": "ollama",
  "embedderModelId": "nomic-embed-text",
  "qdrantUrl": "http://localhost:6333"
}
```

```json
{
  "embedderProvider": "openai",
  "embedderModelId": "text-embedding-3-small",
  "embedderOpenAiApiKey": "sk-your-key",
  "qdrantUrl": "http://localhost:6333"
}
```

```json
{
  "embedderProvider": "openai-compatible",
  "embedderModelId": "text-embedding-3-small",
  "embedderOpenAiCompatibleApiKey": "sk-your-key",
  "embedderOpenAiCompatibleBaseUrl": "https://api.openai.com/v1"
}
```

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

### Start MCP HTTP Server

Source: https://github.com/anrgct/autodev-codebase/blob/master/README.md

Starts the MCP server in HTTP streamable mode, recommended for IDE integration. Specify the port using `--port`.

```bash
codebase index --serve --port=3001
```

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

### Quick Start: Call Graph Analysis Demo

Source: https://github.com/anrgct/autodev-codebase/blob/master/README.md

Demonstrates call graph analysis in demo mode. This command shows the relationships between 'app' and 'addUser' functions.

```bash
# Call graph analysis
codebase call --demo --query="app,addUser"
```

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

### CLI Usage Examples: Summary, Query, and Visualization Modes

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/plans/260117-dependency-cli.md

Illustrates various ways to use the dependency CLI, covering summary output (plain text and JSON), query mode for specific functions, and exporting/opening visualization data.

```bash
✅ codebase call                          # 显示统计概览（tree 格式）
✅ codebase call --json                   # 显示统计概览（JSON 格式，包含示例节点）
✅ codebase call --viz graph.json         # 导出完整可视化数据（Cytoscape.js 格式）
✅ codebase call --open                   # 打开可视化查看器
✅ codebase call --viz graph.json --open  # 导出并打开
```

```bash
✅ codebase call --query "getUser"              # 显示依赖树（tree 格式）
✅ codebase call --query "getUser" --json       # 显示依赖树（JSON 格式）
✅ codebase call --query "getUser,validateUser" # 多函数连接分析
```

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

### Start HTTP MCP Server (Default Port)

Source: https://context7.com/anrgct/autodev-codebase/llms.txt

Starts the Model Context Protocol (MCP) server on the default port, indexing the specified project path. The server exposes search and outline tools for AI IDEs.

```bash
codebase index --serve --path=/my/project
```

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

### Migration Example: Scanner Before and After API Redesign

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/plans/260119-list-files-api-redesign.md

Illustrates the refactoring of a scanner component's file listing logic. The 'Before' example uses the older `listFiles` (likely ripgrep-based) and requires manual filtering. The 'After' examples show how to use the new `listOnlyFiles` convenience method or the structured `listFiles` API for type-safe and cleaner code.

```typescript
const [allPaths, _] = await listFiles(directoryPath, true, 10000, {
  pathUtils: this.deps.pathUtils,
  ripgrepPath: 'rg'  // 🔴 需要 ripgrep
})

// 🔴 需要手动过滤
const filePaths = allPaths.filter((p) => !p.endsWith("/"))
```

```typescript
// 方式 1: 使用便捷方法
const [files, _] = await listOnlyFiles(directoryPath, true, 10000, {
  pathUtils: this.deps.pathUtils,
  workspace: this.deps.workspace,  // ✅ 通过 workspace 获取 ignoreService
  fileSystem: this.deps.fileSystem
})

// 方式 2: 使用结构化 API
const result = await listFiles(directoryPath, {
  recursive: true,
  limit: 10000,
  includeDirectories: false,  // 🔥 明确声明只要文件
  deps: {
    pathUtils: this.deps.pathUtils,
    workspace: this.deps.workspace,
    fileSystem: this.deps.fileSystem
  }
})

const files = result.files  // ✅ 类型安全，无需过滤
```

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

### Install @autodev/codebase CLI

Source: https://github.com/anrgct/autodev-codebase/blob/master/README.md

Install the @autodev/codebase CLI globally and configure it to use Ollama for embeddings. This sets up the default embedder provider and model.

```bash
npm install -g @autodev/codebase
codebase config --set embedderProvider=ollama,embedderModelId=nomic-embed-text
```

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

### Complete Autodev Configuration

Source: https://github.com/anrgct/autodev-codebase/blob/master/CONFIG.md

A comprehensive configuration example including Ollama for embedding and reranking, and summarization with specified batch sizes and retries.

```json
{
  "embedderProvider": "ollama",
  "embedderModelId": "nomic-embed-text",
  "embedderOllamaBaseUrl": "http://localhost:11434",
  "qdrantUrl": "http://localhost:6333",
  "vectorSearchMinScore": 0.3,
  "vectorSearchMaxResults": 20,
  
  "rerankerEnabled": true,
  "rerankerProvider": "ollama",
  "rerankerOllamaModelId": "qwen3-vl:4b-instruct",
  "rerankerOllamaBaseUrl": "http://localhost:11434",
  "rerankerMinScore": 0.5,
  "rerankerBatchSize": 10,
  "rerankerConcurrency": 3,
  "rerankerMaxRetries": 3,
  "rerankerRetryDelayMs": 1000,
  
  "summarizerProvider": "ollama",
  "summarizerOllamaBaseUrl": "http://localhost:11434",
  "summarizerOllamaModelId": "qwen3-vl:4b-instruct",
  "summarizerLanguage": "English",
  "summarizerBatchSize": 2,
  "summarizerConcurrency": 2,
  "summarizerMaxRetries": 3,
  "summarizerRetryDelayMs": 1000
}
```

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

### Start HTTP MCP Server (Custom Port)

Source: https://context7.com/anrgct/autodev-codebase/llms.txt

Starts the MCP server on a custom port (e.g., 3001) for the specified project path. This allows running multiple servers or avoiding port conflicts.

```bash
codebase index --serve --port=3001 --path=/my/project
```

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

### Code Outline with AI Summaries CLI Example

Source: https://github.com/anrgct/autodev-codebase/blob/master/README.md

Generate a structured outline of code files with AI-generated summaries. This command processes 'hello.js' and provides summaries for functions and classes.

```bash
# Code outline with AI summaries - Understand code structure at a glance
╭─ ~/workspace/autodev-codebase 
╰─❯ codebase outline 'hello.js' --demo --summarize
# hello.js (23 lines)
└─ Defines a greeting function that logs a personalized hello message and returns a welcome string. Implements a UserManager class managing an array of users with methods to add users and retrieve the current user list. Exports both components for external use.

   2--5 | function greetUser
   └─ Implements user greeting logic by logging a personalized hello message and returning a welcome message

   7--20 | class UserManager
   └─ Manages user data with methods to add users to a list and retrieve all stored users

   12--15 | method addUser
   └─ Adds a user to the users array and logs a confirmation message with the user's name.
```

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

### Start Stdio Adapter for MCP

Source: https://github.com/anrgct/autodev-codebase/blob/master/README.md

Connects to a running MCP server using the stdio adapter, typically for IDEs that require stdio communication.

```bash
# First start the MCP server in one terminal
codebase index --serve --port=3001

# Then connect via stdio adapter in another terminal (for IDEs that require stdio)
codebase stdio --server-url=http://localhost:3001/mcp
```

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

### Install Ollama and Nomic Embed Text

Source: https://github.com/anrgct/autodev-codebase/blob/master/README.md

Install necessary dependencies for local embedding and reranking using Ollama. Ensure Ollama is running before pulling models.

```bash
brew install ollama ripgrep
ollama serve
ollama pull nomic-embed-text
```

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

### OpenAI Setup Command

Source: https://github.com/anrgct/autodev-codebase/blob/master/CONFIG.md

Command to set the OpenAI API key and configure Autodev to use OpenAI for embeddings.

```bash
export OPENAI_API_KEY="sk-your-key"
codebase config --set embedderProvider=openai,embedderModelId=text-embedding-3-small
```

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

### Initialize and Index Codebase

Source: https://github.com/anrgct/autodev-codebase/blob/master/CLAUDE.md

Use CodeIndexManager for initializing the system and starting the code indexing process. Ensure dependencies are correctly created for Node.js environments.

```typescript
import { CodeIndexManager, createNodeDependencies } from './src/index.ts';

const deps = createNodeDependencies();
const manager = CodeIndexManager.getInstance(deps);
await manager.initialize();
await manager.startIndexing();
const results = await manager.searchIndex(query, { limit: 20 });
```

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

### Codebase Index Serve

Source: https://context7.com/anrgct/autodev-codebase/llms.txt

Starts an HTTP Model Context Protocol (MCP) server that exposes `search_codebase` and `outline_codebase` tools to AI IDEs.

```APIDOC
## `codebase index --serve` — HTTP MCP Server

Starts an HTTP Model Context Protocol server that exposes `search_codebase` and `outline_codebase` tools to AI IDEs (Cursor, Windsurf, Claude, etc.). The server uses the Streamable HTTP transport with session management and supports Server-Sent Events. It begins indexing the workspace as soon as it starts and keeps the index up to date via the file watcher.

```bash
# Start the MCP server on the default port
codebase index --serve --path=/my/project

# Custom port
codebase index --serve --port=3001 --path=/my/project

# Output:
# 🔍 Codebase MCP Server running at http://localhost:3001
# 📁 Workspace: /my/project
# 🌐 MCP endpoint: http://localhost:3001/mcp
# ❤️  Health check: http://localhost:3001/health

# Health check
curl http://localhost:3001/health
# {"status":"healthy","timestamp":"2025-01-22T10:00:00.000Z","workspace":"/my/project"}

# IDE configuration (HTTP Streamable — recommended)
# Add to your IDE's MCP config file:
cat << 'EOF'
{
  "mcpServers": {
    "codebase": {
      "url": "http://localhost:3001/mcp"
    }
  }
}
EOF

# Call search_codebase tool via MCP JSON-RPC
curl -X POST http://localhost:3001/mcp \
  -H "Content-Type: application/json" \
  -d '{ 
    "jsonrpc": "2.0",
    "id": 1,
    "method": "tools/call",
    "params": {
      "name": "search_codebase",
      "arguments": {
        "query": "How does user authentication work?",
        "limit": 10,
        "filters": {
          "pathFilters": ["src/**/*.ts", "!**/*.test.ts"],
          "minScore": 0.6
        }
      }
    }
  }'

# Call outline_codebase tool
curl -X POST http://localhost:3001/mcp \
  -H "Content-Type: application/json" \
  -H "MCP-Session-ID: <session-id-from-init>" \
  -d '{ 
    "jsonrpc": "2.0",
    "id": 2,
    "method": "tools/call",
    "params": {
      "name": "outline_codebase",
      "arguments": {
        "path": "src/auth/service.ts",
        "summarize": true,
        "title": false
      }
    }
  }'
```
```

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

### Get Configuration as JSON

Source: https://github.com/anrgct/autodev-codebase/blob/master/CONFIG.md

Obtain the complete configuration hierarchy in JSON format. This is ideal for scripting and programmatic access to configuration settings.

```bash
codebase config --get --json
```

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

### Example of Concise Path Handling in `call.ts`

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/plans/260122-unify-wasm-path-resolution.md

Illustrates an effective approach to path resolution by using only two paths based on the development environment. This method is clear, efficient, and aligns with project structure.

```typescript
const isDevelopment = currentFilePath.endsWith('.ts');
const viewerPath = isDevelopment
  ? path.join(currentDir, '../../static/graph_viewer.html')  // src/commands -> static
  : path.join(currentDir, 'static/graph_viewer.html');       // dist -> dist/static
```

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

### OpenAI-Compatible Summarizer Configuration JSON

Source: https://github.com/anrgct/autodev-codebase/blob/master/CONFIG.md

Example JSON configuration for an OpenAI-compatible summarizer, specifying API base URL, model ID, and API key.

```json
{
  "summarizerProvider": "openai-compatible",
  "summarizerOpenAiCompatibleBaseUrl": "https://api.deepseek.com/v1",
  "summarizerOpenAiCompatibleModelId": "deepseek-chat",
  "summarizerOpenAiCompatibleApiKey": "sk-your-deepseek-key",
  "summarizerLanguage": "English",
  "summarizerBatchSize": 2,
  "summarizerConcurrency": 2,
  "summarizerMaxRetries": 3,
  "summarizerRetryDelayMs": 1000
}
```

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

### Serve MCP Server with Autodev Codebase CLI

Source: https://github.com/anrgct/autodev-codebase/blob/master/README.md

Start an MCP server for code intelligence. Supports HTTP mode on a specified port or a stdio adapter for communication.

```bash
# HTTP mode (recommended)
codebase index --serve --port=3001 --path=/my/project

# Stdio adapter
codebase stdio --server-url=http://localhost:3001/mcp
```

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

### Configuration Management CLI

Source: https://github.com/anrgct/autodev-codebase/blob/master/CLAUDE.md

Manage Autodev configurations, including getting all settings, specific items, and setting project or global configurations.

```bash
# 配置管理
codebase config --get                           # 查看所有配置层
codebase config --get embedderProvider          # 查看特定配置项
codebase config --set embedderProvider=ollama   # 设置项目配置
codebase config --set key=value --global        # 设置全局配置
```

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

### Ollama Summarizer Configuration JSON

Source: https://github.com/anrgct/autodev-codebase/blob/master/CONFIG.md

Example JSON configuration for the Ollama summarizer provider, including base URL, model ID, and other settings.

```json
{
  "summarizerProvider": "ollama",
  "summarizerOllamaBaseUrl": "http://localhost:11434",
  "summarizerOllamaModelId": "qwen3-vl:4b-instruct",
  "summarizerLanguage": "English",
  "summarizerBatchSize": 2,
  "summarizerConcurrency": 2,
  "summarizerMaxRetries": 3,
  "summarizerRetryDelayMs": 1000
}
```

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

### Call Graph Analysis CLI Example

Source: https://github.com/anrgct/autodev-codebase/blob/master/README.md

Analyze function call relationships and execution paths. This command traces calls related to 'app' and 'addUser'.

```bash
# Call graph analysis - Trace function call relationships and execution paths
╭─ ~/workspace/autodev-codebase 
╰─❯ codebase call --demo --query="app,addUser"
Connections between app, addUser:

Found 2 matching node(s):
  - demo/app:L1-29
  - demo/hello.UserManager.addUser:L12-15

Direct connections:
  - demo/app:L1-29 → demo/hello.UserManager.addUser:L12-15

Chains found:
  - demo/app:L1-29 → demo/hello.UserManager.addUser:L12-15
```

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

### Code Index Manager Example

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/project-outline.md

Shows how to use the Code Index Manager for indexing and retrieving code-related information. This is useful for projects requiring code navigation or analysis.

```typescript
function codeIndexManagerExample() {
  // ... implementation details ...
}
```

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

### Inefficient Dual Querying

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/plans/260119-list-files-api-redesign.md

This example illustrates the current implementation's inefficiency, which involves two separate I/O operations: one using ripgrep and another using fs.readdir. This leads to higher overhead and complexity.

```typescript
// 当前实现（ripgrep 时代）
const files = await listFilesWithRipgrep(...)          // 查询 1: ripgrep 子进程
const directories = await listFilteredDirectories(...) // 查询 2: fs.readdir
return formatAndCombineResults(files, directories, limit)
```

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

### Semantic Code Search CLI Example

Source: https://github.com/anrgct/autodev-codebase/blob/master/README.md

Perform semantic code search using natural language queries. This command finds code snippets related to 'user manage'.

```bash
# Semantic code search - Find code by meaning, not just keywords
╭─ ~/workspace/autodev-codebase 
╰─❯ codebase search "user manage" --demo
Found 20 results in 5 files for: "user manage"

==================================================
File: "hello.js"
==================================================
< class UserManager > (L7-20)
class UserManager {
  constructor() {
    this.users = [];
  }

  addUser(user) {
    this.users.push(user);
    console.log('User added:', user.name);
  }

  getUsers() {
    return this.users;
  }
}
……
```

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

### Get Language Configuration Function

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/project-outline.md

Retrieves the configuration object for a specific programming language. This configuration is used to guide the parsing process.

```typescript
function getLanguageConfig(language: string): LanguageConfig | undefined {
  // ... implementation details ...
}
```

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

### Install @rollup/plugin-replace

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/plans/260122-unify-wasm-path-resolution.md

Install the necessary plugin for replacing code during the Rollup build process.

```bash
npm install --save-dev @rollup/plugin-replace
```

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

### Invalid Score Range Example

Source: https://github.com/anrgct/autodev-codebase/blob/master/CONFIG.md

Example of an invalid command attempting to set the vector search minimum score outside the allowed range (0.0 to 1.0).

```bash
# Invalid score range
codebase config --set vectorSearchMinScore=1.5
# Error: Search minimum score must be between 0 and 1
```

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

### Demonstrate File System Interaction

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/project-outline.md

Illustrates how the project interacts with the file system, potentially for reading, writing, or traversing files and directories.

```typescript
function demonstrateFileSystem() {
  // ... implementation details ...
}
```

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

### Reranker Prompt Example

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/07-search.md

Example prompt for evaluating code snippet relevance against a user query. It specifies the desired output format as a JSON array of scores.

```plaintext
请评估以下代码片段与用户查询的相关性。

查询: {用户查询}

候选代码片段:
1. {代码内容1}
2. {代码内容2}
...

请为每个候选片段打分(0-10分)，10分表示完全相关。
以 JSON 数组格式返回分数: [8, 5, 9, ...]
```

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

### View Configuration

Source: https://github.com/anrgct/autodev-codebase/blob/master/README.md

Use these commands to view the current configuration settings. The `--json` flag provides output in JSON format.

```bash
codebase config --get
```

```bash
codebase config --get --json
```

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

### Use Custom Config File Path

Source: https://context7.com/anrgct/autodev-codebase/llms.txt

Demonstrates how to specify a custom configuration file path when running codebase commands for configuration retrieval and indexing.

```bash
codebase --config=/custom/path/config.json config --get
```

```bash
codebase --config=/custom/path/config.json index
```

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

### Advanced Node.js Usage Example

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/project-outline.md

Illustrates more complex scenarios and advanced features of the autodev library. This example may involve multiple configurations or intricate data flows.

```typescript
function advancedUsageExample() {
  // ... implementation details ...
}
```

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

### Use Custom Configuration File

Source: https://github.com/anrgct/autodev-codebase/blob/master/README.md

Specify a custom configuration file using the `--config` flag. This allows for project-specific settings.

```bash
codebase --config=/path/to/config.json config --get
```

```bash
codebase --config=/path/to/config.json config --set embedderProvider=ollama
```

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

### Build and Run Commands

Source: https://github.com/anrgct/autodev-codebase/blob/master/AGENTS.md

Standard npm scripts for building, type checking, development mode, running the MCP server, and executing tests.

```bash
npm run build
npm run type-check
npm run dev
npm run mcp-server
npm run test
npm run test:e2e
```

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

### Build and Run Commands

Source: https://github.com/anrgct/autodev-codebase/blob/master/CLAUDE.md

Standard npm scripts for building, type checking, development mode, running the MCP server, and executing tests.

```bash
npm run build          # 构建
npm run type-check     # 类型检查
npm run dev            # 用 demo 目录的开发模式
npm run mcp-server     # 启动 MCP 服务器（端口 3001）
npm run test           # vitest 单元测试
npm run test:e2e       # e2e 测试
```

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

### View Configuration Hierarchy

Source: https://github.com/anrgct/autodev-codebase/blob/master/CONFIG.md

Use this command to view the complete configuration hierarchy. It helps in understanding how different configuration sources are layered and prioritized.

```bash
codebase config --get
```

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

### Get Language Configs Function

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/project-outline.md

Retrieves configurations for all supported languages.

```typescript
function getLanguageConfigs(): LanguageConfig[] {
  // ... implementation details ...
}
```

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

### Get Supported Languages Function

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/project-outline.md

Returns a list of programming languages that the parser supports.

```typescript
function getSupportedLanguages(): string[] {
  // ... implementation details ...
}
```

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

### Test Case for analyzeConnections

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/plans/260117-dependency-cli.md

Example test case in `call.test.ts` demonstrating the addition of the `depth` parameter to `analyzeConnections` calls.

```typescript
// src/commands/__tests__/call.test.ts
// 所有 analyzeConnections 调用都添加 depth 参数
analyzeConnections(result.nodes, 'functionA,functionB', 10)
```

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

### NodeWorkspace Get Glob Ignore Patterns

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/04-ignore.md

Retrieves glob patterns used for ignore filtering within the NodeWorkspace context.

```typescript
// NodeWorkspace.getGlobIgnorePatterns:54
// getGlobIgnorePatterns() (NodeWorkspace.getGlobIgnorePatterns:54)
```

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

### Use Custom Configuration File

Source: https://github.com/anrgct/autodev-codebase/blob/master/CONFIG.md

Specify a custom path for the configuration file when running commands. This allows for flexible configuration management.

```bash
codebase --config=/path/to/config.json config --get
```

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

### Configure Ollama Summarizer

Source: https://github.com/anrgct/autodev-codebase/blob/master/CONFIG.md

Set up the CLI to use Ollama for code summarization, specifying the model and other parameters.

```bash
codebase config --set summarizerProvider=ollama,summarizerOllamaModelId=qwen3-vl:4b-instruct
```

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

### View Specific Configuration Item

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/05-config.md

Use this command to retrieve the value of a particular configuration setting. Ensure you specify the exact configuration key.

```bash
codebase config --get embedderProvider vectorSearchMinScore
```

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

### Configuration Priority Levels

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/architecture.md

Details the four-tiered configuration priority system, from CLI arguments to built-in defaults, showing how settings are overridden.

```text-chart
[配置优先级] (从高到低)
1. CLI参数 ─── 运行时覆盖
   └── --path, --log-level, --force等
   
2. 项目配置 ─── ./autodev-config.json
   └── 项目级持久化设置
   
3. 全局配置 ─── ~/.autodev-cache/autodev-config.json
   └── 用户级默认设置
   
4. 内置默认值 ─── 代码中的默认配置
```

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

### IDE Configuration for Stdio Adapter

Source: https://github.com/anrgct/autodev-codebase/blob/master/README.md

JSON configuration for IDEs to use the codebase command with stdio arguments to connect to an MCP server.

```json
{
  "mcpServers": {
    "codebase": {
      "command": "codebase",
      "args": ["stdio", "--server-url=http://localhost:3001/mcp"]
    }
  }
}
```

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

### Create Module Node on Analysis Start

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/plans/260121-top-level-calls-not-tracked.md

This modification adds a call to `createModuleNode` at the beginning of the `analyze` method to establish a node representing the module itself.

```typescript
public async analyze(): Promise<ParseOutput> {
  // 创建 module 节点
  this.createModuleNode()
  
  const tree = this.parser.parse(this.content)
  // ... 现有逻辑
}
```

```typescript
private createModuleNode(): void {
  const moduleId = this.getModulePath()  // 复用现有方法，保持 ID 格式一致
  const moduleNode: DependencyNode = {
    id: moduleId,
    name: path.basename(this.filePath),
    componentType: 'module',
    filePath: this.filePath,
    relativePath: this.getRelativePath(),
    startLine: 1,
    endLine: this.lines.length,
    dependsOn: new Set(),
    language: this.nodeTypes.extensions.values().next().value,
  }
  this.nodes.set(moduleId, moduleNode)
}
```

```typescript
private getModuleNodeId(): string {
  return this.getModulePath()  // 直接使用现有的 getModulePath() 方法
}
```

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

### Codebase Index Initialization Flow

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/01-index.md

The initialization process involves creating Node.js platform dependencies and instantiating the CodeIndexManager.

```text-chart
[Initialization Flow] (Creating and configuring CodeIndexManager)
indexHandler:232
  ↓
initializeManager:118
  ↓
createDependencies:77 → createNodeDependencies:29
  ↓
CodeIndexManager.getInstance:42
  ↓
CodeIndexManager.initialize:124
```

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

### Manage configuration settings

Source: https://context7.com/anrgct/autodev-codebase/llms.txt

Use `codebase config` to read and write configuration. Supports viewing all or specific keys, setting project-level or global values, and JSON output. Project-level settings are saved to `./autodev-config.json`.

```bash
# View all configuration layers
codebase config --get
```

```bash
# View specific keys
codebase config --get embedderProvider qdrantUrl
```

```bash
# JSON output for scripting
codebase config --get --json
```

```bash
# Set project-level config (writes to ./autodev-config.json)
codebase config --set embedderProvider=ollama,embedderModelId=nomic-embed-text
```

```bash
# Set global config
codebase config --set --global qdrantUrl=http://localhost:6333
```

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

### Get Type Label

Source: https://github.com/anrgct/autodev-codebase/blob/master/static/graph_viewer.html

Returns a localized or abbreviated label for a given node type. Used for displaying type information in a user-friendly format.

```javascript
function getTypeLabel(type) {
    const labels = {
        'class': currentLang === 'en' ? 'Class' : '类',
        'function': currentLang === 'en' ? 'Func' : '函数',
        'module': currentLang === 'en' ? 'Mod' : '模块'
    };
    return labels[type] || type.substring(0, 1).toUpperCase();
}
```

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

### Get Leaf Nodes Function

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/project-outline.md

Identifies and returns the leaf nodes of the dependency graph. Leaf nodes are typically modules that do not have any outgoing dependencies.

```typescript
function getLeafNodes(graph: Map<string, string[]>): string[] {
  // ... implementation details ...
}
```

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

### Set Project Configuration

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/05-config.md

This command allows you to set multiple configuration options for the current project. Values are comma-separated.

```bash
codebase config --set embedderProvider=ollama,qdrantUrl=http://localhost:6333
```

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

### Test Full Parsing Functionality

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/project-outline.md

This test function verifies the complete parsing pipeline of the project. It ensures that files are parsed accurately from start to finish.

```typescript
function testFullParsing() {
  // ... implementation details ...
}
```

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

### Generate Code Outline with Summaries

Source: https://github.com/anrgct/autodev-codebase/blob/master/CONFIG.md

Use the 'codebase outline' command with the --summarize flag to generate an intelligent code outline with AI-generated summaries.

```bash
# Generate intelligent code outline with AI summaries
codebase outline "src/**/*.ts" --summarize
```

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

### Remove Pre-creation of Module Nodes

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/plans/260121-top-level-calls-not-tracked.md

Delete the logic that pre-creates module nodes at the start of analysis to avoid redundant nodes for files without dependencies.

```typescript
// ❌ 删除：在 analyze() 开始时创建 module 节点
async analyze(): Promise<ParseOutput> {
  // this.createModuleNode()  // ← 删除这一行
  
  const tree = this.parser.parse(this.content)
  // ...
}
```

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

### Set API Keys via Environment Variables

Source: https://github.com/anrgct/autodev-codebase/blob/master/CONFIG.md

Configure API keys for services like OpenAI, Qdrant, and Gemini using environment variables. This is a common practice for managing sensitive credentials.

```bash
export OPENAI_API_KEY="sk-your-key"
export QDRANT_API_KEY="your-qdrant-key"
export GEMINI_API_KEY="your-gemini-key"
```

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

### Start Indexing in CodeIndexManager

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/01-index.md

The `startIndexing` method in `CodeIndexManager` checks the current status, attempts recovery if in an error state, and then delegates the indexing process to the `CodeIndexOrchestrator`.

```typescript
public async startIndexing(force?: boolean): Promise<void> {
    // Check for error status and attempt recovery
    const currentStatus = this.getCurrentStatus()
    if (currentStatus.systemStatus === "Error") {
        await this.recoverFromError()
        return
    }

    this.assertInitialized()
    await this._orchestrator!.startIndexing(force)
}
```

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

### Demonstrate Search Functionality

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/project-outline.md

Illustrates how to perform searches using the autodev system. This function is useful for understanding the search capabilities and how to query the indexed data.

```typescript
function demonstrateSearch() {
  // ... implementation details ...
}
```

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

### Indexing Data Flow

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/architecture.md

Visualizes the end-to-end data flow for indexing, starting from the file system and progressing through scanning, parsing, embedding, and storing vectors.

```text-chart
[索引数据流] (从文件到向量的完整流程)
文件系统
  ↓
DirectoryScanner ──→ 扫描文件列表
  ↓
IgnoreService ──→ 应用忽略规则
  ↓
FileWatcher ──→ 监听文件变化
  ↓
CodeParser ──→ 解析代码结构
  ↓
代码块提取 ──→ 函数、类、方法
  ↓
Embedder ──→ 生成向量嵌入
  ↓
QdrantVectorStore ──→ 存储向量
```

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

### API Documentation: DependencyNode componentType

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/plans/260121-top-level-calls-not-tracked.md

This documentation update introduces the 'module' componentType for DependencyNode, explaining its purpose in tracking top-level call dependencies and providing an example.

```markdown
## DependencyNode

### componentType

节点类型：
- `function` - 函数定义
- `class` - 类定义
- `method` - 方法定义
- `module` - 模块/文件本身（用于追踪顶层调用）

#### module 节点说明

每个分析的文件都会自动创建一个 `module` 节点，用于追踪文件顶层代码的依赖关系。

**示例：**
```javascript
// app.js
const { greetUser } = require('./hello');
greetUser('Alice');  // 顶层调用

// 依赖图：
// app (module) → hello.greetUser
```

**注意：** 内置函数调用（如 `console.log`）会被自动过滤。
```

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

### Run Unit Tests

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/plans/260122-unify-wasm-path-resolution.md

Execute the project's unit tests to verify the changes. Use the `--silent=false` flag for detailed output.

```bash
npm run test -- --silent=false
```

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

### Orchestrator.startIndexing Decision Flow

Source: https://github.com/anrgct/autodev-codebase/blob/master/docs/01-index.md

The orchestrator decides between incremental or full scanning based on the existence of data in the vector store, initializing the vector store and starting a watcher.

```text-chart
[orchestrator.startIndexing:142 Decision Flow] (Incremental vs. Full Scan)
orchestrator.startIndexing:142
  ↓
vectorStore.initialize()
  ↓
Check hasExistingData
├── Yes → Incremental Scan
│   ├── Mark indexing incomplete
│   ├── scanner.scanDirectory:119 (Incremental)
│   ├── _startWatcher:86 (Start watching)
│   └── Mark indexing complete
└── No → Full Scan
    ├── Mark indexing incomplete
    ├── scanner.scanDirectory:119 (Full)
    ├── _startWatcher:86 (Start watching)
    └── Mark indexing complete
```

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

### Codebase CLI: Configuration Management

Source: https://github.com/anrgct/autodev-codebase/blob/master/AGENTS.md

Manage Autodev configuration settings. View all configuration layers, get specific settings, and set project-level or global configurations.

```bash
# Configuration management
codebase config --get                           # View all configuration layers
codebase config --get embedderProvider          # View a specific configuration item
codebase config --set embedderProvider=ollama   # Set project configuration
codebase config --set key=value --global        # Set global configuration
```

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

### Set Qdrant Docker Environment

Source: https://github.com/anrgct/autodev-codebase/blob/master/CONFIG.md

Run Qdrant using Docker. This command maps the necessary ports for Qdrant to be accessible.

```bash
docker run -p 6333:6333 -p 6334:6334 qdrant/qdrant
```

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

### MCP Tool: Semantic Code Search Payload

Source: https://github.com/anrgct/autodev-codebase/blob/master/AGENTS.md

Example JSON payload for the `search_codebase` MCP tool, specifying the search query, result limit, and filtering criteria.

```json
{
  "query": "user authentication logic",
  "limit": 20,
  "filters": {
    "pathFilters": ["src/**/*.ts"],
    "minScore": 0.3
  }
}
```