jiazhizhong/higress
1
0
Fork 0
mirror of https://github.com/alibaba/higress.git synced 2026-05-26 13:47:27 +08:00
Code Issues Packages Projects Releases Wiki Activity

[feature] add checking of maliciousUrl & modelHallucination, and adjust consumer specific configs (#3024)

Browse Source
This commit is contained in:
rinfx
2025-10-28 14:12:54 +08:00
committed by GitHub
parent 2076ded06f
commit 2a320f87a6
4 changed files with 365 additions and 105 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
plugins/wasm-go/extensions/ai-security-guard/README.md
View File

@@ -37,8 +37,9 @@ description: 阿里云内容安全检测
| `sensitiveDataLevelBar` | string | optional | S4 | 敏感内容检测拦截风险等级，取值为 `S4`, `S3`, `S2` or `S1` |
| `timeout` | int | optional | 2000 | 调用内容安全服务时的超时时间 |
| `bufferLimit` | int | optional | 1000 | 调用内容安全服务时每段文本的长度限制 |
| `consumerSpecificRequestCheckService` | map | optional | - | 为不同消费者指定特定的请求检测服务 |
| `consumerSpecificResponseCheckService` | map | optional | - | 为不同消费者指定特定的响应检测服务 |
| `consumerRequestCheckService` | map | optional | - | 为不同消费者指定特定的请求检测服务 |
| `consumerResponseCheckService` | map | optional | - | 为不同消费者指定特定的响应检测服务 |
| `consumerRiskLevel` | map | optional | - | 为不同消费者指定各维度的拦截风险等级 |
补充说明一下 `denyMessage`，对非法请求的处理逻辑为：
- 如果配置了 `denyMessage`，返回内容为 `denyMessage` 配置内容，格式为openai格式的流式/非流式响应
@@ -70,6 +71,20 @@ description: 阿里云内容安全检测
![](https://img.alicdn.com/imgextra/i4/O1CN013AbDcn1slCY19inU2_!!6000000005806-0-tps-1754-1320.jpg)
阿里云内容安全配置示例：
```yaml
requestCheckService: llm_query_moderation
responseCheckService: llm_response_moderation
```
阿里云AI安全护栏配置示例：
```yaml
requestCheckService: query_security_check
responseCheckService: response_security_check
```
### 检测输入内容是否合规
```yaml