AI Blocks › Keep your code secure
AI agent attack tester
Give it your AI agent. Get a safe, authorized test plan and a report of where it can be tricked.
AI agent attack tester is a free AI skill for the job "Keep your code secure". Give the instructions to an AI assistant that supports the required files and tools, then review its output.
Free · No account · MIT licence · Made by Burhan Sebin · Level: expert · Thinking: high
Made by Burhan Sebin. See the original on GitHub; we copied it as it is, with its licence.
What it does
A safe, approved plan to find where your AI agent can be tricked, plus a report of what was found. It treats written permission as step one. Without clear scope and approval, it stops at a plan and probes nothing live. First it collects the owner, the exact systems in scope, test accounts, time window, limits and a stop signal. Then it maps what the agent can touch and lists likely attacks, such as injected instructions, too much power, memory poisoning and tricks between agents. Tests use fake identities and planted marker values, never real data. Each case gets an ID and a record. Findings are merged, ranked, and retested after fixes. Cleanup is checked at the end.
You give it
- Written approval and the owner's name
- The agent's tools, access and memory
- The systems and data in scope
- Time window and limits for testing
- Codebase (optional)
You get
- Rules of engagement
- A map of what the agent can reach
- A numbered test plan using fake data
- Findings with proof, impact and fixes
- Retest and cleanup results
Use it when
- You are about to launch an agent that can send, buy or delete things.
- Your boss or client asks if the agent can be talked into doing something it should not.
- You want rules of engagement written before any security test.
- You fixed a weakness and need to prove it is closed.
How to use it
- Add it to your AI once: download it and add the folder to Claude, ChatGPT or Codex as a skill.
- Ask for the job in your own words. Your AI opens the skill when it fits.
- It works best when your AI can run code and read your project files.