Protect AI · open-source
LLM Guard
guardrail-testing
What it does
Input and output scanner library for LLM applications. Bundles PII detection, prompt-injection heuristics, toxicity scoring, secrets detection, code-execution detection, and similar checks that can be composed into pipelines.
What it doesn't do
It's a building-block library, not a turnkey policy engine. Some scanners are noisy; expect tuning. Performance overhead at scale needs design consideration.
Best for
- Building input/output filtering pipelines for in-house LLM apps
- Replacing ad-hoc regex scanning with composable, tested scanners
Not for
- Use-cases where latency budgets are extremely tight without careful scanner selection
Pricing
Free.
Integration effort
medium
Team skill required
Python; comfort tuning thresholds.
Last reviewed: 2026-05-01