Prompt Library
Agents
Prompt Injection Audit
Attack your own agent before someone else does.
The Prompt
My agent's system prompt: [paste] It can access: [tools/data]. Attack it: 1. Write 5 injection attempts a malicious user could send — direct, indirect (via retrieved content), and encoding tricks 2. For each: what damage if it works 3. The minimal system-prompt changes that block each class of attack
Fill in before using
[paste][tools/data]How to use it
- → Paste the prompt into your model of choice — it works with any frontier model.
- → Replace every bracketed placeholder with your specifics. Concrete inputs beat vague ones every time.
- → If the model asks a clarifying question, answer it — these prompts are designed to invite that.
- → Iterate on the output, not the prompt. The structure is already tuned.
securityred-teaming