OWASP
LLM01.
OWASP LLM01 (Prompt Injection) is the primary vulnerability where external input overrides an LLM's system-level instructions. In agentic systems, this risk is magnified because the hijacked instructions can trigger external tool calls, leading to unauthorized data access or system manipulation. SupraWall's platform directly mitigates LLM01 by decoupling instruction processing from action execution.
| What | Answer |
|---|---|
| OWASP Classification | LLM01: Prompt Injection. |
| Vulnerability Level | Critical / Top 1. |
| Attack Mechanism | Input-driven instruction override. |
| Agentic Impact | Unauthorized Tool Execution / Privilege Escalation. |
| Defense Strategy | Deterministic Execution Guards (SupraWall). |
Beyond Simple Filters
Action-Centric Defense
OWASP recommends defensive coding and sanitization, but LLMs are stochastic. SupraWall adds a deterministic layer that validates the ACTION, not just the intent.
Full Standard Alignment
SupraWall helps you satisfy OWASP LLM Top 10 requirements for both LLM01 (Prompt Injection) and LLM02 (Insecure Output Handling) in autonomous systems.
Align with Global Standards.
Don't leave your agent security to chance. Implement the deterministic firewall that satisfies OWASP and enterprise security bars.
Join the Private Beta