安装方式
手动下载安装
下载 ZIP 后解压到技能目录即可安装。若在桌面客户端 WebView中直接下载出现异常,本站会改为提示页 + 原始链接,请按页内说明操作。
下载 ZIP (shub-claude-code-runner-v1.0.0.zip)触发指令
/claude-code-runner
跨平台安装指引
该技能声明兼容以下 1 个平台,将 ZIP 解压到对应目录即可被识别。
unzip shub-claude-code-runner-v1.0.0.zip -d ~/.claude/skills/
mkdir -p 创建;启用 Skill 后请重启对应 Agent 让配置生效。
使用指南
Claude Code PTY 运行器
围绕 Claude Code PTY 运行器:基于 PTY 在非 TTY 环境调用 Claude Code,自动应答提示并处理文件同步,适合自动化流水线中的编码步骤。 无需在每次任务前把零散英文说明手工拼进上下文,也 减少 与客户端默认行为脱节的试错;具体命令、钩子与 JSON 参数仍以 ZIP 包内 SKILL.md 为权威。下文结构与站内 MCP CLI 类专题稿相同:何时用、前置、流程、速查与故障。
何时使用
- 基于 PTY 在非 TTY 环境调用 Claude Code,自动应答提示并处理文件同步,适合自动化流水线中的编码步骤
- 已获取本技能 ZIP,并准备在 Claude Code / OpenClaw 中按 SKILL.md 挂载。
- 希望用中文专题稿快速判断「该不该启用」,再深入英文 SKILL 查参数与边界。
- 需要与团队对齐同一套触发方式、目录约定或回调格式时。
前置条件
- 通用:可运行 Claude Code 或文档要求的客户端;有可读写的项目工作区(或 SKILL.md 指定的沙箱目录)。
- 权威细节:API Key / OAuth、钩子路径、环境变量以 ZIP 内 SKILL.md 为准。
- PTY:非 TTY 环境调用时,伪终端与提示应答策略按 SKILL.md(避免卡在读秒确认)。
典型流程
- 从 ClawHub / 站内分发获取技能 ZIP,校验版本与校验和(若提供)。
- 阅读 SKILL.md 的安装段落:目录落点、客户端类型(Claude Code / OpenClaw / 脚本)。
- 用文档中的最小示例完成第一次调用(单文件修改、单次查询或单次委派)。
- 确认工作目录、权限边界与输出路径后,再处理多文件或长耗时任务。
- 需要回调 / Webhook / 通知时,按 SKILL.md 配置端点并在测试环境先验通。
与 ZIP / SKILL.md 的关系
站内专题稿与 MCP CLI 类 oss 稿同样:概括何时用、怎么接、怎么排错;命令模板、钩子名、JSON 字段、版本矩阵一律以 ZIP 内 SKILL.md 与 ClawHub 上游为准。
命令示例(摘自包内 SKILL.md)
以下为从上游 SKILL.md(或入库正文)自动抽取的终端/脚本片段;路径、环境变量与参数以当前 ZIP 与官方说明为准。
ClawHub slug:claude-code-runner(安装命令以 SKILL.md / claw CLI 为准)。
# Clone the skill
git clone https://github.com/lhl09120/claude-code-runner-en.git
# Make script executable
chmod +x claude-code-runner-en/scripts/run_claude.py
python3 scripts/run_claude.py /path/to/project "Your task description here"
which claude
站内入库时的触发命令(完整语义见 ZIP):
# 使用本技能时可在对话中引用或执行上述指令;完整参数与示例见下载包内 SKILL.md。
/claude-code-runner
最佳实践
- 先 SKILL.md 再猜参数;站内专题稿不替代 schema 与必填字段说明。
- 委派任务时写清验收标准(命令、文件路径、测试命令),减少来回追问。
- 长任务用文档推荐的回调 / 日志落盘代替高频轮询,省 Token 也省机器负载。
- 多技能同时启用时,注意钩子加载顺序与重复工具调用(以 SKILL.md 冲突说明为准)。
调试与排错
- 打开 stderr 与客户端日志;PTY/tmux 场景同时看面板最后几十行输出。
- 参数错误时对照 SKILL.md 中的 JSON/CLI 示例(引号、转义、工作目录)。
- 网络类失败:查代理、防火墙、MCP 传输方式(stdio / HTTP / SSE)。
速查
| 动作 | 说明 |
|------|------|
| 获取技能包 | ClawHub / 站内 ZIP,核对版本 |
| 权威步骤 | 优先阅读 ZIP 内 SKILL.md |
| 首次试跑 | 使用 SKILL.md 最小示例 |
| 验收 | 对照路径、测试命令或回调负载 |
常见故障
- 无输出或立即退出 → 工作目录错误、依赖未装、或 Claude Code 未登录;按 SKILL.md 自检清单执行。
- 权限被拒绝 → 检查沙箱路径、
--permission-mode与工具白名单。 - 与简介不符 → 以英文 SKILL 与上游仓库为准,站内稿仅作结构化导读。
- 卡在交互提示 → PTY 应答规则或未使用非交互 flag;见 SKILL.md。
# Claude Code Runner
## Overview
A wrapper skill for running Claude Code programmatically in non-interactive environments. Uses PTY (pseudo-terminal) to handle TTY-required operations and automatically responds to confirmation prompts.
## Features
- **PTY-based execution**: Works in non-TTY environments (containers, CI/CD, background processes)
- **Auto-respond to prompts**: Automatically answers "Do you want to..." confirmations
- **User switching**: Runs as specified non-root user
- **File synchronization**: Copies project to temp directory, executes, syncs changes back
- **Timeout handling**: Configurable timeout with proper cleanup
- **Output capture**: Captures and returns full stdout/stderr
## Installation
```bash
# Clone the skill
git clone https://github.com/lhl09120/claude-code-runner-en.git
# Make script executable
chmod +x claude-code-runner-en/scripts/run_claude.py
```
## Usage
### Basic Usage
```python
from claude_code_runner import run_claude_code
result = run_claude_code(
workdir='/path/to/project',
prompt='Refactor the authentication module to use JWT tokens',
user='lighthouse',
timeout=300
)
print(result)
```
### Via Command Line
```bash
python3 scripts/run_claude.py /path/to/project "Your task description here"
```
### Advanced Options
```python
result = run_claude_code(
workdir='/root/repo/my-project',
prompt='''
1. Review the codebase
2. Identify security vulnerabilities
3. Fix any issues found
4. Add appropriate tests
''',
user='developer',
timeout=600 # 10 minutes
)
```
## API Reference
### `run_claude_code(workdir, prompt, user='lighthouse', timeout=300)`
Execute a Claude Code task in a PTY environment.
**Parameters:**
- `workdir` (str): Working directory containing the project
- `prompt` (str): Natural language task description
- `user` (str): User to run as (default: 'lighthouse')
- `timeout` (int): Timeout in seconds (default: 300)
**Returns:**
- `str`: Combined stdout and stderr output
**Behavior:**
1. Copies project to temporary directory
2. Changes ownership to specified user
3. Executes Claude Code via PTY
4. Auto-responds to confirmation prompts
5. Syncs changes back to original directory
6. Cleans up temporary files
## Use Cases
### 1. Automated Code Review
```python
result = run_claude_code(
workdir='/root/repo/project',
prompt='Review this codebase and identify potential bugs or improvements'
)
```
### 2. Refactoring Tasks
```python
result = run_claude_code(
workdir='/root/repo/legacy-app',
prompt='Refactor the database layer to use SQLAlchemy ORM instead of raw SQL'
)
```
### 3. Adding Features
```python
result = run_claude_code(
workdir='/root/repo/api-service',
prompt='''
Add a new REST endpoint for user profile management:
- GET /api/users/{id}/profile
- PUT /api/users/{id}/profile
- Include validation and error handling
- Add unit tests
'''
)
```
### 4. Bug Fixes
```python
result = run_claude_code(
workdir='/root/repo/web-app',
prompt='Fix the memory leak in the WebSocket connection handler'
)
```
## Requirements
- Python 3.8+
- Claude Code installed and in PATH
- Unix-like environment (Linux/macOS)
- Root or sudo access (for user switching)
## Configuration
### Environment Variables
- `CLAUDE_CODE_USER`: Default user to run as (default: 'lighthouse')
- `CLAUDE_CODE_TIMEOUT`: Default timeout in seconds (default: 300)
### Customization
Edit `scripts/run_claude.py` to customize:
- Auto-response keywords
- Temp directory location
- Sync behavior
- Output formatting
## Troubleshooting
### "Permission denied" errors
Ensure the script is run with sufficient privileges to:
- Create temporary directories
- Change file ownership
- Switch to target user
### Claude Code not found
Make sure Claude Code is installed and in the system PATH:
```bash
which claude
```
### Task timeout
Increase the timeout for long-running tasks:
```python
run_claude_code(workdir, prompt, timeout=600) # 10 minutes
```
### Interactive prompts not auto-responded
Add new prompt patterns to the auto-respond logic:
```python
if b'new prompt text' in output:
os.write(master_fd, b'y\n')
```
## Limitations
- Requires Unix-like environment (uses PTY)
- Requires root/sudo for user switching
- Claude Code must be installed separately
- May not handle all edge cases of interactive prompts
## License
MIT License
Copyright (c) 2026 lhl09120
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.
See LICENSE file for full details.
## Changelog
### v1.0.0 (2026-02-27)
- Initial release
- PTY-based Claude Code execution
- Auto-response to confirmation prompts
- File synchronization
- User switching support