Free · No install · API key optional
System Prompt Analyzer
Paste your agent's system prompt below to see which OWASP LLM Top 10 security controls it addresses — and exactly what's missing.
Try a sample:
Pattern matching misses controls phrased differently from templates.
Semantic analysis uses
Your API key is used only in your browser to call Anthropic's API directly. It is never sent to our servers and is not stored anywhere.
claude-haiku to evaluate each control contextually —
catching things like "politely decline" or "you only answer HR questions."Your API key is used only in your browser to call Anthropic's API directly. It is never sent to our servers and is not stored anywhere.
0 characters
0/8
controls detected
Security control checks
Run these checks from your terminal:
For dynamic attack testing:
pip install checkagent then
checkagent analyze-prompt your_prompt.txtFor dynamic attack testing:
checkagent scan module:agent_fn
— runs 101 probes including prompt injection, PII leakage, and jailbreak attempts.