Phantom - Autonomous Offensive Security Intelligence AI-po

项目说明

<di="center">

<br/>

☠ PHANTOM

Autonomous Ad Simulation Platform

*AI-nati testing — autonomous reconnaissance, exploitation, and results.*

<br/>

![PyPI]()
![Python]()

<br/>

Quick Start &nbsp;·&nbsp; Architecture &nbsp;·&nbsp; Usage &nbsp;·&nbsp; Configuration &nbsp;·&nbsp; Contributing

<br/>
</div>

---

O

Phantom is an autonomous AI penetration testing agent built on the ReAct (Reason–Act) loop. It connects a large language model to over 30 professional security tools, runs all offensi inside an isolated Docker sandbox, and produces reports — entirely without human inter.

Unlike C scanners, Phantom reasons about your target: it reads HTTP responses, forms hypotheses, selects the right tool, chains multi-step exploits, then writes and executes a proof-of-concept script to confirm every finding before it appears in a report.

| | Traditional Scanners | Phantom |
|--|--|--|
| Approach | Signature matching against C | LLM reasoning + adaptive tool chaining |
| False Positives | 40–70% — requires manual triage | Every finding with a working PoC |
| Depth | Single-pass HTTP probe | Multi-phase: recon → exploit → |
| Adaptability | Fixed rules, static payloads | Adapts to target responses in real time |
| No | Known CVEs only | Logic flaws + no paths |
| Reporting | Generic lists | MITRE ATT&CK mapped, compliance-ready |

---

Core Capabilities

<table>
<tr>
<td align="center">🧠</td>
<td><strong>Autonomous ReAct Loop</strong> — Plans, executes tools, reads results, re-plans. Handles dead ends and unexpected responses without human guidance.</td>
</tr>
<tr>
<td align="center">🔧</td>
<td><strong>53 Security Tools</strong> — nmap · nuclei · sqlmap · ffuf · httpx · katana · subfinder · nikto · gobuster · arjun · semgrep · playwright — all orchestrated automatically.</td>
</tr>
<tr>
<td align="center">🐳</td>
<td><strong>Ephemeral Docker Sandbox</strong> — All offensi runs in a network-restricted Kali Linux container. Zero host filesystem access. Container is destroyed after every scan.</td>
</tr>
<tr>
<td align="center">⚡</td>
<td><strong>Multi-Agent Parallelism</strong> — Spawns specialized sub-agents (SQLi, XSS, recon) that work concurrently and report findings to the coordinator.</td>
</tr>
<tr>
<td align="center">🛡️</td>
<td><strong>7-Layer Defense Model</strong> — Scope guard → Tool firewall → Docker sandbox → Cost limiter → Time budget → HMAC audit trail → Output sanitizer.</td>
</tr>
<tr>
<td align="center">✅</td>
<td><strong> Findings Only</strong> — No hallucinations. Every reported includes raw HTTP e, reproduction steps, and a working exploit script.</td>
</tr>
<tr>
<td align="center">🗺️</td>
<td><strong>MITRE ATT&CK Enrichment</strong> — Automatic CWE, CAPEC, technique-le, and CVSS 3.1 scoring per finding.</td>
</tr>
<tr>
<td align="center">📋</td>
<td><strong>Compliance Co</strong> — OWASP Top 10 (2021) · PCI DSS v4.0 · NIST 800-53 — mapped automatically per finding.</td>
</tr>
<tr>
<td align="center">💾</td>
<td><strong>Knowledge Persistence</strong> — Cross-scan memory stores hosts, past findings, and false-positi. Each scan learns from the last.</td>
</tr>
<tr>
<td align="center">💰</td>
<td><strong>Full Cost Control</strong> — Per-request and per-scan budget caps. Every token and every dollar tracked in real time.</td>
</tr>
</table>

---

Architecture

<details open>
<summary><strong>① System Architecture — Component O</strong></summary>
<br/>

</details>

<details>
<summary><strong>② Scan Execution Flow — Phase by Phase</strong></summary>
<br/>

</details>

<details>
<summary><strong>③ Agent ReAct Loop — Decision Cycle</strong></summary>
<br/>

</details>

<details>
<summary><strong>④ Docker Sandbox — Isolation Architecture</strong></summary>
<br/>

</details>

<details>
<summary><strong>⑤ 7-Layer Defense Model — Request Lifecycle</strong></summary>
<br/>

</details>

---

Quick Start

Requirements: Docker · Python 3.12+ · An LLM API key

> First run pulls the sandbox image (~13 GB). This happens once. Subsequent scans start in under 10 seconds.

---

Usage

Scan Profiles

| Profile | Max Iterations | Typical Duration | Best For |
|---------|:--------------:|:----------------:|----------|
| quick | 300 | ~15–60 min | CI/CD gates, rapid triage |
| standard | 120 | ~20–45 min | Regular security testing |
| deep | 300 | 1–3 hours | Full audits, compliance (default) |
| stealth | 60 | ~30–60 min | Covert assessments, WAF-aware targets |
| api_only | 100 | ~20–45 min | REST/GraphQL API-focused scans |

Output

Every scan produces:

Post-Scan Enrichment Pipeline

Every scan automatically runs a 7-stage enrichment pass:

| Stage | Action |
|-------|--------|
| 1. MITRE ATT&CK | CWE, CAPEC, technique-le |
| 2. Compliance | OWASP Top 10 · PCI DSS v4 · NIST 800-53 |
| 3. Attack Graph | Dependency-based path analysis |
| 4. Nuclei Templates | Auto-generated YAML for regression testing |
| 5. Knowledge Store | Persistent cross-scan memory updated |
| 6. Notifications | Webhook / Slack alerts for critical findings |
| 7. Reports | JSON + HTML + Markdown output |

---

CI/CD Integration

---

Configuration

<details>
<summary><strong>En </strong></summary>

| | Description | Default |
|----------|-------------|---------|
| PHANTOM_LLM | LLM model (LiteLLM format) | openai/gpt-4o |
| LLM_API_KEY | API key — comma-separated for rotation | — |
| PHANTOM_REASONING_EFFORT | low / medium / high | medium |
| PHANTOM_SCAN_MODE | Default scan profile | standard |
| PHANTOM_IMAGE | Sandbox Docker image | ghcr.io/usta0x001/phantom-sandbox:latest |
| PHANTOM_MAX_COST | Hard stop when total scan cost (USD) reaches this limit | — |
| PHANTOM_PER_REQUEST_CEILING | Hard stop when a single LLM call exceeds this cost in USD | — |
| LLM_MAX_TOKENS | O max output tokens per LLM call (o scan-mode defaults: quick=4000, stealth=6000, default=8000) | — |
| PHANTOM_WEBHOOK_URL | Webhook URL for critical alerts | — |
| PHANTOM_DISABLE_BROWSER | Disable Playwright browser | false |
| PHANTOM_TELEMETRY | Enable anonymous usage telemetry | false |

</details>

<details>
<summary><strong>Supported LLM Pro</strong></summary>

Phantom uses LiteLLM — 100+ pro work out of the box:

| Provider | Example Model | Notes |
|----------|--------------|-------|
| OpenAI | openai/gpt-4o | Best o quality |
| Anthropic | anthropic/claude-opus-4-5 | Strong multi-step reasoning |
| Google | gemini/gemini-2.5-pro | Huge context window |
| Groq | groq/llama-3.3-70b- | Free tier, very fast |
| DeepSeek | deepseek/deepseek-chat | Excellent cost efficiency |
| OpenRouter | openrouter/deepseek/deepseek-v3.2 | Multi-provider routing |
| Ollama | ollama/llama3.1 | Fully local — no API key required |
| Azure OpenAI | azure/gpt-4o | Enterprise deployments |

</details>

---

Security Audit

Phantom has undergone extensi auditing across multiple :

| Se | Identified | Fixed |
|----------|:----------:|:-----:|
| Critical | 8 | 8 |
| High | 19 | 19 |
| Medium | 34 | 34 |
| Low | 27 | 27 |
| Total | 88 | 88 |

All 88 identified issues are resolved. See CHANGELOG.md for the full history.

---

Testing

See tests/ for the test suite. Integration and end-to-end tests require a li en.

---

Documentation

| Resource | Description |
|----------|-------------|
| Architecture | Deep-di design |
| Documentation | Full API and configuration reference |
| Contributing | De guidelines |
| Changelog | history and release notes |

---

Docker Sandbox — Setup Guide

Phantom runs all offensi inside an isolated Docker sandbox container. This section covers setup for fresh installs and custom en.

Default Sandbox Image

The default image is ghcr.io/usta0x001/phantom-sandbox:latest — a pre-built Kali Linux container with all security tools installed.

Requirements:

Using a Custom Sandbox Image

O the image or config:

Air-Gapped / Offline En

If your en has no internet access:

Sandbox Is Working

---

Contributing

Contributions are welcome. See CONTRIBUTING.md for setup instructions.

---

License

Apache License 2.0 — see LICENSE.

---

Acknowledgements

Built on the shoulders of giants:

LiteLLM · Nuclei · SQLMap · Playwright · Textual · Rich · ffuf · Subfinder · Caido

---

<di="center">

☠ PHANTOM &nbsp;—&nbsp; *Autonomous Ad Simulation Platform*

PyPI &nbsp;·&nbsp; GitHub &nbsp;·&nbsp; Issues &nbsp;·&nbsp; Made by Usta0x001

<br/>

> ⚠ Legal Notice: Only test systems you own or ha written authorization to test.
> Unauthorized access to computer systems is illegal. The authors assume no liability for misuse.

</div>

项目信息

开发语言：Python

Star 数：16

技术要点

该项目属于安全软件方向的开源资源，代码结构清晰，适合安全研究人员学习参考。项目围绕 Autonomous Offensive Security Intelligence AI-powered multi-agent penetration testing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。