Safe Config Modifier

安全修改 OpenClaw 配置文件。用于任何需要修改 ~/.openclaw/openclaw.json 的场景，包括：模型切换、channel 配置、tools 配置、skill 安装等。确保修改前备份、预览（脱敏 key）、并获得用户确认。

This skill appears coherent for safely editing ~/.openclaw/openclaw.json. Before installing or running it: (1) inspect the bundled scripts locally (they are small); (2) always run the sanitize/preview scripts rather than the raw 'cat' commands in SKILL.md to avoid exposing secrets; (3) keep automatic/autonomous agent execution in mind — ensure the agent only runs this skill when you intend it to; (4) verify the sanitize patterns cover the secrets you actually use (they are simple sed rules and may miss some keys or have minor regex bugs); (5) the curl example will make an outbound request if executed — only run it when you want to test connectivity. If you need higher assurance, run the scripts manually in a safe environment first and ensure backups are taken (the skill also recommends doing so).

Type: OpenClaw Skill Name: safe-config Version: 1.0.0 The 'safe-config' skill is a utility designed to manage OpenClaw configuration files with a strong emphasis on safety and data privacy. It includes scripts (preview.sh, sanitize.sh, validate.sh) that use sed and jq to mask sensitive fields like apiKeys and tokens before displaying them to the user. A notable security feature is the requirement for a specific confirmation phrase ('ojbk可以改了') to prevent the AI agent from performing modifications based on vague or accidental user input. The code performs standard file operations (cp, mv, jq) and lacks any indicators of data exfiltration, persistence, or malicious execution.