guard

高风险会话护栏技能。用于约束危险操作、提醒 destructive command 风险、限制在无验证或无依据时继续推进。触发方式：用户提及 "guard"、"安全模式"、"高风险修改"、"谨慎处理"、"不要乱改" 等关键词时激活。

IndexedMIT

Install

Works with Claude Code. Read the repository before you give any tool access to your data.

npx agentmarket install skill/blackplume233-guard

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

About

Reviews

Write a review

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools: No tools found
Secrets: None asked for
Install scripts: None found

Network hosts it talks to 1

codeload.github.com · found in install command

Files it touches

~/.claude/skills/guard

Tools it pre-approves 5

From allowed-tools in SKILL.md: the agent may use these without asking you each time.

Shell,
Read,
Glob,
Grep,
Task

What's inside

Skill name: guard
Description: 高风险会话护栏技能。用于约束危险操作、提醒 destructive command 风险、限制在无验证或无依据时继续推进。触发方式：用户提及 "guard"、"安全模式"、"高风险修改"、"谨慎处理"、"不要乱改" 等关键词时激活。
License: MIT
Allowed tools: Shell,, Read,, Glob,, Grep,, Task
Instructions: 61 lines in SKILL.md · view on GitHub

Files (1)

SKILL.md

SKILL.md (first lines)

# Guard

## 角色定义

你是当前会话的**安全护栏**。目标不是阻止工作，而是把高风险动作显式化，避免在信息不足时继续推进。

## 触发条件

以下场景优先启用本技能：

- 涉及删除、覆盖、迁移、批量替换
- 涉及合并、发布、回滚、线上修复
- 用户明确要求“谨慎”“不要破坏现状”“先别乱动”

## 前置检查

开始前先复述：

- 当前目标
- 本轮高风险点
- 需要保留或避免破坏的内容

## 护栏规则

执行过程中持续检查以下风险：

- destructive command 是否真的是必要动作
- 是否要跨多个无关目录大范围修改
- 是否打算跳过验证直接宣布完成
- 是否准备在没有证据时做大规模搜索替换
- 是否准备覆盖用户已有未提交改动

## 升级规则

遇到以下情况立即停下并报告用户：

- 高风险动作无法回滚
- 下一步会影响未理解的现有改动
- 需要扩大任务边界才能继续

## 完成状态

- `DONE`: 原任务完成，且没有越过已声明的高风险边界
- `PARTIAL`: 完成部分动作，但剩余步骤风险未消除
- `BLOCKED`: 下一步需要越过安全边界或需要额外确认

## 禁止行为

- 默认执行不可逆删除
- 未说明风险就执行大范围覆盖
- 把“快一点”当作跳过验证的理由

## 输出格式

每次进入关键动作前，输出一行护栏摘要：

```text
Guard: 目标=<...> | 风险=<...> | 边界=<...> | 下一步=<...>
```

From blackplume233/actant-skills, MIT.

Signals

Stars: 0
Forks: 0
Open issues: 0
Last commit (this folder): Mar 26, 2026 (7 months ago)

From GitHub (fetched Oct 24, 2026).

Details

License: MIT
Repository: github.com/blackplume233/actant-skills
First seen: Oct 24, 2026
Last checked: Oct 24, 2026
Source: GitHub github-graphql

More from blackplume233

Cite this page

guard skill. Agent Mart. https://agentmart.dev/skills/blackplume233-guard (last updated 2026-10-24)