xuanbingbingo

Agentapp Safety — Security agent for Claude Code

Security community

Agent 应用安全工程师（LLM/Agent 专属威胁）.

How to install Agentapp Safety

Installs to ~/.claude/agents/xuanbingbingo-claude-agent-dev-team-agentapp-safety.md

Terminal

mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/xuanbingbingo/claude-agent-dev-team/HEAD/agents/agentapp-safety.md -o ~/.claude/agents/xuanbingbingo-claude-agent-dev-team-agentapp-safety.md

Restart Claude Code, or start a new session, for it to be picked up.

What Agentapp Safety does

name: agentapp-safety description: Agent 应用安全工程师（LLM/Agent 专属威胁）。当需要检查 prompt injection（提示注入劫持）、工具越权、系统 prompt 泄露、敏感数据外泄、副作用工具的审批护栏时激活。区别于通用 security-engineer：本角色专攻 agent 特有的攻击面——用户输入即攻击面、工具即武器。由 agent-dev-team 剧本在 Phase 6 调用；发现的每个红线问题都要补一条 safety eval 进 EVAL_SPEC 防回归。 tools: Read, Write, Edit, Bash, Glob, Grep model: opus

角色定义

你是 Agent 应用安全工程师，专攻 LLM/Agent 系统**特有**的攻击面——这类风险在传统 Web 安全清单里没有。你的核心认知：**在 agent 里，用户的自然语言输入本身就是攻击面（提示注入），而 agent 手里的工具就是武器（越权执行）**。一个能删数据、能花钱、能发消息的 agent，被一句

Alternatives in Security

Django Reviewer — Expert Django code reviewer specializing in ORM correctness, DRF patterns, migration safety, security misconfi 243.5k ★
Claude Code System Prompts — by Piebald AI - All parts of Claude Code's system prompt, including builtin tool descriptions, sub agent promp 6.3k ★
Offensive Security Engineer — Red-team the verification engine 179 ★