What is PromptGuard?
PromptGuard is an adaptive security platform designed to identify and prevent prompt injection attacks targeting Large Language Model applications.
The Security Boundary
PromptGuard works as a security boundary between incoming user prompts and downstream LLM providers. Each prompt is inspected before it is forwarded to an AI model.
The platform analyses instruction manipulation, suspicious behaviour, encoded payloads, role manipulation and attempts to access sensitive information.
→ (BLOCK) → [0 Tokens Dispatched to Model]
The Prompt Injection Problem
Modern AI applications concatenate untrusted user inputs with system prompts, developer guardrails, and retrieved content into a single conversational context window.
Because language models natively struggle to differentiate between instructions to be executed and data to be processed, adversaries can craft prompts that override developer directives, reveal confidential tokens, or execute arbitrary tool functions.
Zero-Trust Pre-Execution Guarantees
- Zero LLM Token Leakage
Prompts classified as critical threats are dropped immediately at the network perimeter before dispatch.
- Deterministic Clamped Scoring
Heuristic weights strictly produce a bounded 0 to 100 risk index with deterministic explanations.
- Local Simulation Ready
Audit defense mechanisms locally on your machine without transmitting proprietary prompts to third-party clouds.
- Open Source Integrity
Fully auditable Python codebase distributed under the permissive MIT Open Source License.