Getting Started

Welcome to Agentic Security! This guide will help you orient yourself within the project and start using the tool quickly.

Project Overview

Agentic Security is an open-source vulnerability scanner for Large Language Models (LLMs). It provides both a command line interface and a FastAPI server so you can probe models for weaknesses such as jailbreaks or refusal patterns. The tool supports multimodal attacks, multi-step scans and reinforcement-learning based probes.

Repository Layout

agentic_security/
├── __main__.py    - CLI entry point
├── app.py         - FastAPI app assembly
├── lib.py         - SecurityScanner and utilities
├── config.py      - Configuration handling
├── core/          - app state and logging helpers
├── probe_actor/   - scanning logic and RL modules
├── probe_data/    - dataset registry and loaders
├── routes/        - API endpoints
└── ui/            - Web UI assets (Vue)

tests/ contains unit tests, and docs/ houses the project documentation.

Quick Start

Ensure you have completed the installation steps.
Run the following command to start the application:
```
agentic_security
```
The server will start on http://localhost:8718.
Explore available commands with:
```
agentic_security --help
```

Basic Usage

To view available commands, run:
```
agentic_security --help
```

Next Steps

Review the Quickstart Guide for a fast setup walkthrough.
Check http_spec.md to learn how LLM endpoints are described.
Browse the probe_actor and probe_data modules to understand how scanning works and how datasets are loaded.
Explore the Configuration section to customize your setup.
Run the tests in tests/ to verify your environment once dependencies are installed.

This guide should give you a solid foundation for exploring and extending Agentic Security. For more details, see the rest of the documentation.

2.0 KiB Raw Blame History