OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Security & Safety

Fuzzyai

FuzzyAI is an Apache-2.0 command-line tool that fuzzes language-model APIs with generated and adversarial inputs to help developers and security researchers identify jailbreaks and potential vulnerabilities.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Fuzzyai — OpenViST project profile in AI Security & Safety.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use FuzzyAI when you need a command-line workflow for testing an LLM API with adversarial inputs and reviewing its responses for jailbreaks or other potential security weaknesses.

Who it is for

  • Developers evaluating security risks in LLM applications and API integrations.
  • Security researchers conducting red-team assessments of language-model APIs.

Documented uses

  • Probe an LLM API with attack strategies and test prompts.
  • Assess responses to individual prompts or a prompt file.
  • Test a REST API using a raw HTTP request file and endpoint settings.
  • Use an auxiliary model to classify responses in a fuzzing example.
Before you choose

Requirements & limits

The repository declares an Apache-2.0 license. Installation instructions support pip from the Git repository or a local clone with Poetry. The README also describes optional local use with Ollama and shows provider examples that use API credentials.

  • FuzzyAI is a CLI, with an experimental Web UI offered as another interface after installation.
  • The documented setup paths are pip installation from Git or editable installation from a clone using Poetry.
  • Ollama with a local model is optional; examples also show OpenAI, Anthropic and REST API use.
  • The README describes attack modes, prompt inputs, REST request configuration and response classification through an auxiliary model.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install FuzzyAI and open the fuzzer help

Set up the CLI from an existing Python project and view its fuzz command options.

Before you start

  • An existing Python project
  • A package manager that can install from a Git repository

Steps

  1. Install FuzzyAI with the command below.

    Documented command
    pip install git+https://github.com/cyberark/FuzzyAI.git

  2. Run fuzzyai fuzz -h to display the fuzzer help.

Expected outcome

The FuzzyAI CLI is installed and its fuzz command help is displayed.

Guide limitations

  • This setup guide does not configure a model or run a fuzzing assessment.
Documentation 1 ↗
Source-based practical guide

Set up the experimental Web UI

Install FuzzyAI from the repository and start its experimental Web UI.

Before you start

  • A local repository clone
  • Poetry

Steps

  1. Clone https://github.com/cyberark/FuzzyAI.git and change into the FuzzyAI directory.

  2. Install the project with poetry run pip install -e ..

  3. Start the interface with poetry run fuzzyai webui; use --port 9000 to specify port 9000.

Expected outcome

The experimental Web UI is started using the selected port.

Guide limitations

  • The README labels the Web UI experimental.
  • The guide covers starting the interface, not configuring or conducting a fuzzing assessment.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
11 October 2026
Original website title
cyberark/FuzzyAI · Official homepage title · retained in its source language
How we review sourcesSuggest a correction