https://github.com/Tencent/AI-Infra-Guard
AI infrastructure security assessment tool designed to discover and detect potential security risks in AI systems.
https://github.com/Tencent/AI-Infra-Guard
Last synced: about 1 month ago
JSON representation
AI infrastructure security assessment tool designed to discover and detect potential security risks in AI systems.
- Host: GitHub
- URL: https://github.com/Tencent/AI-Infra-Guard
- Owner: Tencent
- License: other
- Created: 2024-12-25T06:39:30.000Z (4 months ago)
- Default Branch: main
- Last Pushed: 2025-03-27T12:36:23.000Z (about 1 month ago)
- Last Synced: 2025-03-27T13:35:11.156Z (about 1 month ago)
- Language: Go
- Homepage:
- Size: 11.3 MB
- Stars: 890
- Watchers: 18
- Forks: 92
- Open Issues: 8
-
Metadata Files:
- Readme: README.md
- License: License.txt
Awesome Lists containing this project
- awesome-hacking-lists - Tencent/AI-Infra-Guard - AI infrastructure security assessment tool designed to discover and detect potential security risks in AI systems. (Go)
README
# π‘οΈ AI Infrastructure Security Assessment System
[δΈζη](./README_CN.md)
AI Infra Guard is an **efficient**, **lightweight**, and **user-friendly** AI infrastructure security assessment tool designed to discover and detect potential security risks in AI systems.
## π Quick Preview
**WEBUI**
**Command line**
## π Project Highlights
* **Efficient Scanning*** Supports 28 AI framework fingerprint identifications
* Covers 200+ security vulnerability databases
* **Easy to Use**
* Ready to use out of the box, no complex configuration
* Fingerprint and vulnerability YAML rule definitions
* Flexible matching syntax
* **Lightweight**
* Clean and efficient core components
* Small binary size, low resource usage
* Cross-platform support
## π AI Component Coverage
| Component Name | Vulnerability Count |
|---------------------|---------------------|
| anythingllm | 8 |
| langchain | 33 |
| Chuanhugpt | 0 |
| clickhouse | 22 |
| comfy_mtb | 1 |
| ComfyUI-Prompt-Preview | 1 |
| ComfyUI-Custom-Scripts | 1 |
| comfyui | 1 |
| dify | 11 |
| fastchat-webui | 0 |
| fastchat | 1 |
| feast | 0 |
| gradio | 42 |
| jupyterlab | 6 |
| jupyter-notebook | 1 |
| jupyter-server | 13 |
| kubeflow | 4 |
| kubepi | 5 |
| llamafactory | 1 |
| llmstudio | 0 |
| ollama | 7 |
| open-webui | 8 |
| pyload-ng | 18 |
| qanything | 2 |
| ragflow | 2 |
| ray | 4 |
| tensorboard | 0 |
| vllm | 4 |
| xinference | 0 |
| triton-inference-server | 7 |
## π¦ Installation and Usage
### Installation
Download from: [https://github.com/Tencent/AI-Infra-Guard/releases](https://github.com/Tencent/AI-Infra-Guard/releases)
### Usage
**WEBUI**
```
./ai-infra-guard -ws
```
**Local Scan**
```
./ai-infra-guard -localscan
```
**Single Target**
```
./ai-infra-guard -target [IP/Domain]
```
**Multiple Targets**
```
./ai-infra-guard -target [IP/Domain] -target [IP/Domain]
```
**Read from File**
```
./ai-infra-guard -file target.txt
```
**AI Analysis**
```
# hunyuan token
./ai-infra-guard -target [IP/Domain] -ai -hunyuan-token [Hunyuan token]
# deepseek token
./ai-infra-guard -target [IP/Domain] -ai -deepseek-token [deepseek token]
```
## π Fingerprint Matching Rules
AI Infra Guard is based on WEB fingerprint identification components. Fingerprint rules are in the [data/fingerprints](./data/fingerprints) directory, and vulnerability matching rules are in the [data/vuln](./data/vuln) directory.
### Example: Gradio Fingerprint Rule
```yaml
info:
name: gradio
author: Security Team
severity: info
metadata:
product: gradio
vendor: gradio
http:
- method: GET
path: '/'
matchers:
- body="window.gradio_config = {" || body="document.getElementsByTagName(\"gradio-app\");"
```
### π οΈ Fingerprint Matching Syntax
#### Matching Locations
- Title
- Body
- Header
- Icon hash
#### Logical Operators
- `=` Fuzzy match
- `==` Exact match
- `!=` Not equal
- `~=` Regular expression match
- `&&` AND
- `||` OR
- `()` Grouping
## π€ Contributing
Community contributions are highly welcome! Feel free to submit [PRs](https://github.com/Tencent/AI-Infra-Guard/pulls) and [Issues](https://github.com/Tencent/AI-Infra-Guard/issues).
## Star History
[](https://star-history.com/#Tencent/AI-Infra-Guard&Date)
## π License
"AI Infra Guard" is licensed under the MIT License except for the third-party components. Visit [License.txt](./License.txt)
---
π‘ **Tip**: Stay updated to protect your AI infrastructure security!