rogue - AI Agent Evaluator & Red Team Platform

项目说明

Rogue — AI Agent E & Red Team Platform

![]()

<di="center">

<a href=" target="_blank"><img src=" alt="qualifire-de | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>

!Tests

<img src="./freddy-rogue.png" width="200"/>

Stress-test your AI agents before attackers do.

Discord Community · Quick Start · Documentation

</div>

---

Two Ways to Harden Your Agent

<table>
<tr>
<td width="50%" ="top">

🎯 Automatic E

Test your agent against business policies and expected behaviors.

Best for: Regression testing, behavior , policy compliance

</td>
<td width="50%" ="top">

🔴 Red Teaming

Simulate ad attacks to find security .

Best for: Security audits, penetration testing, compliance reporting

</td>
</tr>
</table>

---

Architecture

Rogue operates on a client-ser with multiple interfaces:

| Component | Description |
| ---------- | ------------------------------------------- |
| Server | Core e & red team logic |
| TUI | Modern terminal interface (Go + Bubble Tea) |
| CLI | Non-interactive mode for CI/CD pipelines |

Supported Protocols

| Protocol | Transport | Description |
| ---------- | -------------------- | ---------------------------------------------------------------------------------- |
| A2A | HTTP | Google's Agent-to-Agent protocol |
| MCP | SSE, STREAMABLE_HTTP | Model Context Protocol via send_message tool |
| Python | — | Direct Python function calls (no network protocol) |

See examples in examples/ for reference implementations.

#### Python Entrypoint

For agents implemented as Python functions without A2A or MCP:

1. Create a Python file with a call_agent function:

2. Run Rogue with Python protocol:

Or via TUI: select "Python" as the protocol and enter the file path.

See examples/python_entrypoint_stub.py for a complete example.

---

🔥 Quick Start

Prerequisites

Installation

Try It With the Example Agent

Configure in the UI:

---

Running Modes

| Mode | Command | Description |
| ------- | --------------------- | ----------------------- |
| Default | u | Server + TUI |
| Server | u server | Backend only |
| TUI | u tui | Terminal client |
| CLI | u cli | Non-interactive (CI/CD) |

Ser

CLI Options

| Option | Description |
| ------------------------ | ------------------------------------------- |
| --config-file | Path to config JSON |
| --e | Agent endpoint (required) |
| --judge-llm | LLM for e (required) |
| --business-context | Context string or --business-context-file |
| --input-scenarios-file | Scenarios JSON |
| --output-report-file | Report output path |

---

Red Teaming

Scan Types

| Type | | Attacks | Time |
| ---------- | --------------- | ------------- | ---------- |
| Basic | 5 curated | 6 | ~2-3 min |
| Full | 75+ | 40+ | ~30-45 min |
| Custom | User-selected | User-selected | |

Compliance Frameworks

Attack Categories

| Category | Examples |
| ------------------ | --------------------------------------- |
| Encoding | Base64, ROT13, Leetspeak |
| Social Engineering | Roleplay, trust building |
| Injection | Prompt injection, SQL injection |
| Semantic | Goal redirection, context poisoning |
| Technical | Gray-box probing, permission escalation |

Risk Scoring (C)

Each receives a 0-10 risk score based on:

Reproducible Scans

Perfect for regression testing and security fixes.

---

Configuration

En

Config File (.rogue/user_config.json)

---

Key Features

| Feature | Description |
| ------------------------ | -------------------------------------------- |
| 🔄 Dynamic Scenarios | Auto-generate tests from business context |
| 👀 Li | Watch agent con in real-time |
| 📊 Comprehensi | Markdown, CSV, JSON exports |
| 🔍 Multi-Faceted Testing | Policy compliance + security |
| 🤖 Model Support | OpenAI, Anthropic, Google () |
| 🛡️ C | Industry-standard risk assessment |
| 🔁 Reproducible | Deterministic scans with random seeds |

---

Documentation

---

Contributing

1. Fork the repository
2. Create a branch (git checkout -b feature/amazing-feature)
3. Commit changes (git commit -m 'Add amazing feature')
4. Push (git push origin feature/amazing-feature)
5. Open a Pull Request

---

License

Licensed under a proprietary license — see LICENSE.

Free for personal and internal use. Commercial hosting requires licensing.
Contact:

项目信息

开发语言：Python

Star 数：1066

技术要点

该项目属于红蓝攻防方向的开源资源，代码结构清晰，适合安全研究人员学习参考。项目围绕AI Agent Evaluator & Red Team Platform展开，提供了可运行的实现，对理解相关安全机制、工具原理与攻防技术具有较高的研究价值。

从技术实现角度看，此类项目的核心在于对底层原理的深入理解与工程化落地。开发者在研究过程中可以结合官方文档与社区讨论，梳理项目的初始化流程、核心模块职责与数据流转，逐步掌握其中的技术细节与设计思路。

使用与学习建议

1. 下载源码压缩包，解压到本地环境。

2. 根据项目类型准备对应的运行环境与依赖。

3. 按照 README 中的步骤完成配置与部署。

4. 在隔离的测试环境中运行验证，避免影响生产系统。

免责声明

本资源来源于开源社区，仅供安全研究、技术学习与合法授权范围内的攻防演练使用，请勿用于任何非法或侵害他人权益的用途。使用本项目产生的一切后果由使用者自行承担。