AI Agent Security

Secure AI agents against prompt injection, tool abuse, and data exfiltration with defense-in-depth controls.

9 stars 4 forksAI, LLMs & Agents

Install

skills CLI (Claude Code, Cursor, Codex and more)
npx skills add huytran1120/vietnam-agentic-skills --skill ai-agent-security
Manual install
git clone --depth 1 https://github.com/huytran1120/vietnam-agentic-skills.git /tmp/vietnam-agentic-skills
mkdir -p ~/.claude/skills
cp -r /tmp/vietnam-agentic-skills/skills/ai-agent-security ~/.claude/skills/ai-agent-security

Personal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.

From the SKILL.md

# AI Agent Security

Protect agentic AI systems from adversarial input, unsafe tool execution, data leakage, and privilege abuse with layered security controls.

These are partial design sketches, not a working security boundary. Regexes/classifiers have false positives and false negatives; prompt text cannot grant permissions. Enforce tenant identity, resource scope and approved actions in the tool service using authenticated server-side context. Input signing authenticates origin, not the safety of signed content. Bind approvals to the exact operation/arguments and recheck immediately before execution. Test failure paths and concurrency before production use.

## Prerequisites

- Python 3.10+ for guardrail code examples
- Docker or Podman for sandbox execution
- OpenTelemetry collector for audit logging
- Familiarity with your agent framework (LangChain, CrewAI, Autogen, custom)
- Access to policy engine (OPA/Cedar) for permission boundaries

## Threat Model — STRIDE for AI Agents

AI agents introduce a unique threat surface. Apply STRIDE specifically to agentic components:

| Threat | Agent-Specific Example | Control |
|--------|----------------------|---------|
| **Spoofing** | Attacker crafts input that mimics a trusted internal tool response | Signed tool responses, HMAC verification |
| **Tampering** | Prompt injection modifies agent reasoning mid-chain | Input validation, prompt armoring |
Read the full skill on GitHub

Details

Skill name
ai-agent-security
Path
skills/ai-agent-security/SKILL.md
License
MIT
Repo updated
29 October 2026
Copies
Found in 2 GitHub repos

FAQ

What does the AI Agent Security skill do?

Secure AI agents against prompt injection, tool abuse, and data exfiltration with defense-in-depth controls.

How do I install the AI Agent Security skill?

skills CLI (Claude Code, Cursor, Codex and more): npx skills add huytran1120/vietnam-agentic-skills --skill ai-agent-security. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.

Is AI Agent Security popular?

It lives in huytran1120/vietnam-agentic-skills, which has 9 GitHub stars and 4 forks.

Related skills

Browse the most popular Claude skills, all ai, llms & agents skills, or write your own with the skills generator.