safety-guard

GitHub

通过拦截危险命令和限制文件编辑范围,防止生产环境误操作。提供谨慎、冻结及组合三种保护模式,适用于自主代理运行或敏感操作场景,保障系统安全。

data/skills-md/affaan-m/ecc/safety-guard/SKILL.md NeverSight/learn-skills.dev

触发场景

工作于生产系统时 代理全自主运行时 执行迁移或部署等敏感操作时

安装

npx skills add NeverSight/learn-skills.dev --skill safety-guard -g -y
更多选项

非标准路径

npx skills add https://github.com/NeverSight/learn-skills.dev/tree/main/data/skills-md/affaan-m/ecc/safety-guard -g -y

不安装直接使用

npx skills use NeverSight/learn-skills.dev@safety-guard

指定 Agent (Claude Code)

npx skills add NeverSight/learn-skills.dev --skill safety-guard -a claude-code -g -y

安装 repo 全部 skill

npx skills add NeverSight/learn-skills.dev --all -g -y

预览 repo 内 skill

npx skills add NeverSight/learn-skills.dev --list

SKILL.md

Frontmatter
{
    "name": "safety-guard",
    "origin": "ECC",
    "description": "Use this skill to prevent destructive operations when working on production systems or running agents autonomously."
}

Safety Guard — Prevent Destructive Operations

When to Use

  • When working on production systems
  • When agents are running autonomously (full-auto mode)
  • When you want to restrict edits to a specific directory
  • During sensitive operations (migrations, deploys, data changes)

How It Works

Three modes of protection:

Mode 1: Careful Mode

Intercepts destructive commands before execution and warns:

Watched patterns:
- rm -rf (especially /, ~, or project root)
- git push --force
- git reset --hard
- git checkout . (discard all changes)
- DROP TABLE / DROP DATABASE
- docker system prune
- kubectl delete
- chmod 777
- sudo rm
- npm publish (accidental publishes)
- Any command with --no-verify

When detected: shows what the command does, asks for confirmation, suggests safer alternative.

Mode 2: Freeze Mode

Locks file edits to a specific directory tree:

/safety-guard freeze src/components/

Any Write/Edit outside src/components/ is blocked with an explanation. Useful when you want an agent to focus on one area without touching unrelated code.

Mode 3: Guard Mode (Careful + Freeze combined)

Both protections active. Maximum safety for autonomous agents.

/safety-guard guard --dir src/api/ --allow-read-all

Agents can read anything but only write to src/api/. Destructive commands are blocked everywhere.

Unlock

/safety-guard off

Implementation

Uses PreToolUse hooks to intercept Bash, Write, Edit, and MultiEdit tool calls. Checks the command/path against the active rules before allowing execution.

Integration

  • Enable by default for codex -a never sessions
  • Pair with observability risk scoring in ECC 2.0
  • Logs all blocked actions to ~/.claude/safety-guard.log

版本历史

  • e0220ca 当前 2026-07-05 23:48

同 Skill 集合

data/skills-md/00prabalk00/claude-skills/knowledge-base-gap-finder/SKILL.md
data/skills-md/01000001-01001110/agent-jira-skills/jira-agile/SKILL.md
data/skills-md/01000001-01001110/agent-jira-skills/jira-auth/SKILL.md
data/skills-md/01000001-01001110/agent-jira-skills/jira-issues/SKILL.md
data/skills-md/01000001-01001110/agent-jira-skills/jira-project-management/SKILL.md
data/skills-md/01000001-01001110/agent-jira-skills/jira-projects/SKILL.md
data/skills-md/01000001-01001110/agent-jira-skills/jira-safe/SKILL.md
data/skills-md/01000001-01001110/agent-jira-skills/jira-search/SKILL.md
data/skills-md/01000001-01001110/agent-jira-skills/jira-spaces/SKILL.md
data/skills-md/01000001-01001110/agent-jira-skills/jira-transitions/SKILL.md
data/skills-md/0731coderlee-sudo/wechat-publisher/wechat-publisher/SKILL.md
data/skills-md/0froq/skills/conventionalcommits/SKILL.md
data/skills-md/0froq/skills/nuxt/SKILL.md
data/skills-md/0froq/skills/oq/SKILL.md
data/skills-md/0froq/skills/pinia/SKILL.md
data/skills-md/0froq/skills/pnpm/SKILL.md
data/skills-md/0froq/skills/slidev/SKILL.md
data/skills-md/0froq/skills/tsdown/SKILL.md
data/skills-md/0froq/skills/turborepo/SKILL.md
data/skills-md/0froq/skills/unocss/SKILL.md
data/skills-md/0froq/skills/vitepress/SKILL.md
data/skills-md/0froq/skills/vitest/SKILL.md
data/skills-md/0froq/skills/vue-best-practices/SKILL.md
data/skills-md/0froq/skills/vue-router-best-practices/SKILL.md
data/skills-md/0froq/skills/vue-testing-best-practices/SKILL.md
data/skills-md/0froq/skills/vue/SKILL.md
data/skills-md/0froq/skills/vueuse-functions/SKILL.md
data/skills-md/0froq/skills/web-design-guidelines/SKILL.md
data/skills-md/0juano/agent-skills/bondterminal-x402/SKILL.md
data/skills-md/0juano/agent-skills/edgeone-pages-deploy/SKILL.md
data/skills-md/0juano/agent-skills/ley-ar/SKILL.md
data/skills-md/0juano/agent-skills/ticktick/SKILL.md
data/skills-md/0juano/agent-skills/x-image-cards/SKILL.md
data/skills-md/0juano/x-image-cards/x-image-cards/SKILL.md
data/skills-md/0x0funky/agent-sprite-forge/generate2dsprite/SKILL.md
data/skills-md/0x0funky/agent-sprite-forge/video2dsprite/SKILL.md
data/skills-md/0x2e/superpowers/brainstorming/SKILL.md
data/skills-md/0x2e/superpowers/dispatching-parallel-agents/SKILL.md
data/skills-md/0x2e/superpowers/executing-plans/SKILL.md
data/skills-md/0x2e/superpowers/finishing-a-development-branch/SKILL.md
data/skills-md/0x2e/superpowers/receiving-code-review/SKILL.md
data/skills-md/0x2e/superpowers/requesting-code-review/SKILL.md
data/skills-md/0x2e/superpowers/subagent-driven-development/SKILL.md
data/skills-md/0x2e/superpowers/systematic-debugging/SKILL.md
data/skills-md/0x2e/superpowers/test-driven-development/SKILL.md
data/skills-md/0x2e/superpowers/using-git-worktrees/SKILL.md
data/skills-md/0x2e/superpowers/using-superpowers/SKILL.md
data/skills-md/0x2e/superpowers/verification-before-completion/SKILL.md
data/skills-md/0x2e/superpowers/writing-plans/SKILL.md
data/skills-md/0x2e/superpowers/writing-skills/SKILL.md

元信息

文件数
0
版本
f4b4cf1
Hash
50328f1a
收录时间
2026-07-05 23:48

首页 - Wiki
Copyright © 2011-2026 iteam. Current version is 2.155.2. UTC+08:00, 2026-08-10 19:02
浙ICP备14020137号-1 $访客地图$