garak
github.comlibraryself-hosted
garak is a self-hosted library in the Guardrails & Safety layer of the AI stack. Scanner that probes models for known vulnerability classes.
By Lattice editorial · Facts verified · method
- licence
- Apache-2.0
- language
- Python
- cost
- free
- owned by
- Production & Governance
Verified 2026-09build refuses it from 2027-04
Use garak when
Probing your own system for known vulnerability classes.
Skip garak when
You have no remediation path for whatever it finds.
Stops the bad input before it, and the bad output after. — that is what this layer of the stack is answerable for. See the full Guardrails & Safety section for the rest of the options.
Visit github.comCompare with another toolBadge this entry
[](https://lattice.kunjshah.in/guardrails-safety/garak)Quick answers
- Is garak open source?
- Yes. garak is open source under the Apache-2.0 licence. It is free to use.
- What are the alternatives to garak?
- The closest alternatives to garak are PyRIT.
Reading on guardrails
Alternatives to garak
Listed as an alternative to
- PyRITPyRITMicrosoft's red-team tool for generating and scoring attack prompts.Use when Generating and scoring attack prompts systematically.Skip when You are not authorised to test the target.MIT · self-hosted · free
- promptfoopromptfooDeclarative red-teaming and regression tests for prompts and agents.Use when Declarative red-teaming and regression tests that run in CI.Skip when You need a managed UI rather than a test runner.MIT · self-hosted · free
Also in guardrails
- NeMo GuardrailsNeMo GuardrailsProgrammable rails that constrain conversational flow.Use when Constraining conversational flow with programmable rails.Skip when You need deep semantic moderation — it is not a classifier.Apache-2.0 · self-hosted · free
- Guardrails AIGuardrails AIValidators that check model output against a defined schema.Use when Validators that check output against a schema you define.Skip when You need free-form moderation rather than structural checks.Apache-2.0 · self-hosted · free
- Llama GuardLlama GuardSafety classifier for prompt and response moderation.Use when One classifier covering both prompt and response moderation.Skip when You need domain-specific policy enforcement.Llama-3.1-Community · self-hosted · free
- Microsoft PresidioMicrosoft PresidioPII detection and anonymization for text and images.Use when Detecting and anonymising PII in text and images.Skip when Your data is already pseudonymised at source.MIT · self-hosted · free
- Lakera GuardLakera GuardPrompt injection and jailbreak detection at the gateway.Use when Prompt-injection and jailbreak detection at the gateway.Skip when Requests cannot leave your network.proprietary · saas · usage-based
- PyRITPyRITMicrosoft's red-team tool for generating and scoring attack prompts.Use when Generating and scoring attack prompts systematically.Skip when You are not authorised to test the target.MIT · self-hosted · free
- Invariant GuardrailsInvariant GuardrailsGuardrails as code, enforced inline in the call path.Use when Guardrails as code, enforced inline in the call path.Skip when Self-hosting is a hard requirement.proprietary · managed · usage-based