Validated LLM outputs
Framework that validates, corrects and guarantees the structure of model outputs with reusable guards.
The framework for validating LLM outputs: named Guards (PII checks, schema validation, toxicity) that validate/correct/retry model responses in Python. Open-source core with a hub of community guards
Who it's for
Python teams that want validated, structured LLM outputs with retries and guardrails in production.
Official links
Keep exploring
Concepts in the glossary
1 in catalog
Guardrails AI is written in
2 in catalog