AI Agent

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15

# 安装 nvm
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.0/install.sh | bash
source ~/.bashrc
nvm --version # 验证 nvm 安装
nvm install --lts 
node --version 
npm --version  

# 安装 claude-code
npm install -g @anthropic-ai/claude-code
# 验证
which claude
claude --help
# 更新
claude update

1
2
3
4
5
6
7

Your goal is to update any vulnerable dependencies.

Do the following:

1.Run `npm audit` to find vulnerable installed packages in this project
2.Run `npm audit fix` to apply updates
3.Run tests and verify the updates didn't break anything

1
2
3
4
5
6

{
    "permissions": {
        "allow": ["mcp__playwright"],
        "deny": []
    }
}

1
2
3
4

claude mcp list # 查看当前配置的 MCP 服务器
claude mcp add playwright npx @playwright/mcp@latest # 安装 playwright 服务，Playwright 提供控制浏览器的能力
# Open the browser and navigate to localhost:3000
claude mcp remove playwright # 移除配置

1
2
3

# codex cli安装好之后配置mcp
claude mcp add codex -s user -- codex -m gpt-5.1-codex-max -c model_reasoning_effort="high" mcp-server
claude mcp remove codex # 移除配置

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26

{
   "hooks": {
       "PreToolUse": [
           {
               "matcher": "Read|Grep",
               "hooks": [
                {
                    "type": "command",
                    "command": "node ./hooks/read_hook.js"
                }
               ]
           }
       ],
       "PostToolUse": [
           {
               "matcher": "Write|Edit|MultiEdit",
               "hooks": [
                {
                    "type": "command",
                    "command": "true"
                }
               ]
           }
       ]
   } 
}

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17

// ./hooks/read_hook.js
async function main() {
    const chunks = [];
    for await (const chunk of process.stdin) {
        chunks.push(chunk);
    }
    const toolArgs = JSON.parse(Buffer.concat(chunks).toString());
    const readPath =
          toolArgs.tool_input?.file_path || toolArgs.tool_input?.path || "";
    
    // TODO: 让claude不要读隐私文件,比如.env
    if (readPath.includes('.env')) {
        console.error("You cannot read the .env file");
        process.exit(2);
    }
}
main();

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12

import anyio

from claude_code_sdk import query

async def main():
    prompt = "Hello, Claude! Tell me a joke."
    async for message in query(prompt):
        print(message)

anyio.run(main)

# cli: claude -p "Hello, Claude! Tell me a joke."

1
2
3
4
5
6

my-plugin/
├── .claude-plugin/
│   └── plugin.json
└── skills/
    └── code-review/
        └── SKILL.md

1
2
3

/plugin list # 列出插件
/plugin uninstall <plugin-name> # 卸载插件
/plugin marketplace remove <marketplace-name> # 移除市场

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16

---
name: explain-code
description: Explains code with visual diagrams and analogies. Use when explaining how code works, teaching about a codebase, or when the user asks "how does this work?"
disable-model-invocation: true   # 禁止自动调用
allowed-tools: ["Read", "Grep"]  # 限制可用工具
context: fork                    # 在子代理中运行
---

When explaining code, always include:

1. **Start with an analogy**: Compare the code to something from everyday life
2. **Draw a diagram**: Use ASCII art to show the flow, structure, or relationships
3. **Walk through the code**: Explain step-by-step what happens
4. **Highlight a gotcha**: What's a common mistake or misconception?

Keep explanations conversational. For complex concepts, use multiple analogies.

1
2
3
4
5
6
7

#1. 计划模式 进行探索：Claude 阅读文件并回答问题，无需更改文件。
#2.  Create a plan. 按 Ctrl+G 在文本编辑器中打开计划，直接编辑，然后 Claude 继续。
I want to add xxxx. What files need to change? What's the session flow? Create a plan.
#3. 切回普通模式，让claude 编码，并对照计划进行验证。
implement the xxxx from your plan. write tests for the callback handler, run the test suite and fix any failures.
#4. commit提交 
commit with a descriptive message and open a PR

1

npx @jeffreycao/copilot-api@latest start 

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16

"env": {
    "ANTHROPIC_BASE_URL": "http://localhost:4141",
    "ANTHROPIC_AUTH_TOKEN": "dummy",
    "ANTHROPIC_MODEL": "gpt-5.4",
    "ANTHROPIC_DEFAULT_SONNET_MODEL": "gpt-5.4",
    "ANTHROPIC_DEFAULT_HAIKU_MODEL": "gpt-5-mini",
    "DISABLE_NON_ESSENTIAL_MODEL_CALLS": "1",
    "CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC": "1",
    "CLAUDE_CODE_ATTRIBUTION_HEADER": "0",
    "CLAUDE_CODE_ENABLE_PROMPT_SUGGESTION": "false"
  },
  "permissions": {
    "deny": [
      "WebSearch"
    ]
  }

1
2
3
4

# 安装 Codex CLI，并能在终端调用 codex 命令
npm install -g @openai/codex
# 配置~/.codex/auth.json 输入codex进去后填入apikey就会自动生成
# 配置 ~/.codex/config.toml

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11

# ~/.codex/config.toml 使用自己的提供商
model_provider = "codex-for-me"
model = "gpt-5.2-codex"
model_reasoning_effort = "high"
disable_response_storage = true

[model_providers.codex-for-me]
name = "codex-for-me"
base_url = "https://api-vip.codex-for.me/v1"
wire_api = "responses"
requires_openai_auth = true

1
2

npm update -g @openai/codex # 更新
npm uninstall -g @openai/codex # 卸载

1

npm i -g opencode-ai@latest