Prompt Injection
Detects f-string, .format(), %-formatting, and concatenation injection vectors with AST taint tracking.
LLM Armor
Find LLM security vulnerabilities before attackers do. Open-source static analysis scanner covering 7 of 10 OWASP LLM Top 10 risks.
What it detects
Section titled “What it detects”Sensitive Info Disclosure
Finds leaked API keys (OpenAI, Anthropic, Google, HuggingFace) across all file types.
Excessive Agency
Flags wildcard tool access, dangerous tool classes, disabled approval gates, and dynamic dispatch.
Unbounded Consumption
Catches LLM API calls missing max_tokens limits.
Install in 10 seconds
Section titled “Install in 10 seconds”pip install llmarmorllmarmor scan ./your-app/