rembg: remove-image-background

使用 rembg AI 模型去除图片背景，生成透明背景的 PNG 图片。首次使用需运行 setup/install.py 初始化环境。

This skill appears to do what it claims (local background removal) but has several things to consider before installing: - It will create a virtual environment in your home (~/.venv/rembg), modify your shell startup file to add that venv to PATH, and install packages from PyPI. If you prefer not to have persistent changes, run the install steps manually inside an environment you control. - The rembg package will download model files (~176MB) from the network (GitHub releases) on first use. Expect network activity and disk usage. - The scripts embed input file paths inside Python -c code using single-quoted strings. Filenames containing quotes or specially crafted names could break those strings and allow execution of arbitrary Python code. Avoid running these scripts on untrusted inputs, inspect and/or sanitize filenames, or adapt the scripts to pass filenames safely (e.g., via command-line args parsed with argparse or by using subprocess with a safe argument list inside the spawned Python process). - If you are cautious: review setup/install.py and the requirements.txt, consider creating and activating your own venv and running pip install there, or run the scripts in an isolated container or VM first. If you proceed, inspect any edits to ~/.zshrc / ~/.bashrc and remove them if undesired.

Type: OpenClaw Skill Name: rembg Version: 0.0.1 The skill bundle contains a critical Remote Code Execution (RCE) vulnerability in 'scripts/remove_bg.py' and 'scripts/batch_remove_bg.py' where unsanitized file paths are interpolated into a Python command string executed via 'subprocess.run'. Additionally, 'setup/install.py' performs intrusive system modifications by automatically modifying shell configuration files (~/.zshrc, ~/.bashrc) and Windows environment variables to persist PATH changes. While these behaviors are aligned with the stated goal of setting up the 'rembg' tool, the lack of input sanitization and the aggressive persistence mechanisms pose a significant security risk.