Skip to content

Garak

Open-source LLM vulnerability scanner with plugins and prompts that test large language model security.

Security

How it works / How to use

Open-source LLM vulnerability scanner with plugins and prompts that test large language model security.

  1. Open Garak and select the project, alert, or scan target.
  2. Open-source LLM vulnerability scanner with plugins and prompts that test large language model security.
  3. State the risk question and the evidence you already have.
  4. Review findings and confirm remediations before applying changes.

How to prompt

When prompting Garak, name the asset or alert, the risk question, and the output you need (triage, fix, or report). Ask for evidence or confidence when decisions matter.

Review this finding: [summary]. Context: [repo or alert]. Output: severity, why it matters, and the safest next step.

Best output tips

  • Prefer verified findings over volume. Ask what is exploitable or reachable.
  • Confirm fixes against the official docs or advisory before shipping.

Try this AI

Try Garak

Product Details

Pricing, features, limits and latest updates

Garak

Open-source LLM vulnerability scanner with plugins and prompts that test large language model security.

No listed price

Pricing Plans

No listed plans.

Key Features

No listed features.

Ideas / Prompt experiences

Share a prompt that worked for you. Username and email are shown with your submission. External links are not allowed.

No prompt experiences yet. Be the first to share one below.

Share your experience

Required fields are marked. Variation, result, and explanation are optional.

Aikido Security logo

Unified security platform that detects, pentests, and blocks threats from code to runtime.

Security

Finds stolen content, deepfakes, and impersonations and files takedowns.

SecurityPaid

Checkmarx One logo

Agentic application security platform that finds and prioritizes vulnerabilities across code, supply chain, AI components, and runtime.

Security