📦 AI守门人

v1.0.4

LLM API 代理服务管理工具。支持多 Provider 转发（百炼/OpenRouter/NVIDIA）、内容安全审计、健康监控。使用场景：(1)启动/停止/重启代理服务 (2)查看代理状态和统计 (3)配置内容过滤规则。

0· 200·0 当前·0 累计

by @394286006·MIT-0

AI模型访问开发工具 API工具大语言模型

下载技能包

License

MIT-0

最后更新

2026/4/12

安全扫描

VirusTotal

可疑

查看报告

OpenClaw

可疑

medium confidence

该技能基本如其所言（一个带过滤的本地 LLM 代理），但存在需谨慎对待的行为——最显著的是对代理请求进行本地日志记录（可能包含 API 密钥/敏感载荷），以及一个可选的第三层审查功能，若启用可能会发起外部调用。

评估建议

This skill is largely consistent with its advertised purpose, but proceed cautiously. Before installing or running: (1) Inspect the code paths that perform logging and confirm whether Authorization headers or request bodies are sanitized — if not, assume API keys and sensitive payloads will be written to ~/.openclaw/logs/llm-proxy and adjust log permissions/rotation or modify the code to redact sensitive headers. (2) Keep the proxy bound to 127.0.0.1 (do not change listen_host to 0.0.0.0) unless...

详细分析 ▾

✓ 用途与能力

Name/description (LLM proxy + content auditing) match the files and runtime instructions: control scripts, a Python proxy, config and filtering rules. The listed provider endpoints align with a multi-provider proxy. There are no surprising external service credentials requested in metadata.

ℹ 指令范围

SKILL.md and README instruct running the provided scripts (./scripts/llm-proxy-ctl.sh) and editing scripts/llm-proxy-config.json; the code reads that config and writes logs under ~/.openclaw/logs/llm-proxy. That is expected for a proxy. Caveat: the skill appears to log request information (request logs described in README and a LogWriter in code). Logged request entries may include headers or bodies that contain API keys or other secrets unless explicitly sanitized. SKILL.md does not mention redaction. Also the rules include an optional L3 review model (disabled by default) which, if enabled, could cause outbound model calls.

✓ 安装机制

No install spec; this is instruction-only with bundled scripts. Nothing is downloaded from remote hosts during install. The code is included in the skill bundle, so no external install URLs or archive extraction risks are present.

ℹ 凭证需求

The registry metadata declares no required env vars or credentials, which is consistent with a local proxy. However the code does read optional environment variables (LLM_PROXY_CONFIG, LLM_PROXY_PORT, RULES_FILE) and will forward whichever Authorization header the client supplies to upstream providers. Users may therefore send provider API keys through the proxy; combined with local logging, those keys could be stored. No unrelated credentials are requested by the skill itself.

✓ 持久化与权限

always:false and no special platform privileges requested. The skill writes logs and PID files to user-owned locations (~/.openclaw, /tmp) — normal for a local service. It does not request to auto-enable itself or modify other skills.

安全有层次，运行前请审查代码。

License

MIT-0

可自由使用、修改和再分发，无需署名。

查看条款 ↗

运行时依赖

无特殊依赖

版本

latestv1.0.42026/3/18

- 主要文档更新：简化并重组 SKILL.md，提升可读性与用户引导。 - 用单一配置文件和共享脚本替换 _meta.json 及相关参考文档。 - 新增配置文件：scripts/llm-proxy-config.json，实现服务配置集中化与清晰化。 - 新增实用脚本：scripts/llm-proxy-common.sh。 - 明确并分类所有支持的 provider 与安全层。 - 扩充命令行与手动使用、服务日志及高级内容过滤设置的说明。

● 可疑

安装命令

点击复制

官方npx clawhub@latest install llm-proxy

镜像加速npx clawhub@latest install llm-proxy --registry https://cn.longxiaskill.com

技能文档

本地 LLM API 代理服务，提供多 Provider 转发、内容安全审计等功能。

功能特性

🔄 多 Provider 转发 - 百炼、OpenRouter、NVIDIA NIM
🔒 内容安全审计 - 两层审核机制（恶意指令 + 敏感内容）
📊 请求统计 - 实时统计请求数、错误率、告警数
📝 日志记录 - 所有请求 JSONL 格式存档

快速开始

查看代理状态

bash  
llm-proxy-ctl.sh status

  
启动代理

bash  
llm-proxy-ctl.sh start

  
停止代理

bash  
llm-proxy-ctl.sh stop

  
重启代理

bash  
llm-proxy-ctl.sh restart

  
查看实时日志

bash  
llm-proxy-ctl.sh logs

  
Provider 配置  
代理支持以下 Provider 路径前缀：  
| 前缀 | 目标 API |  
|------|----------|  
|

/bailian | 阿里云百炼 https://coding.dashscope.aliyuncs.com/v1

 |  
|

/openrouter | OpenRouter https://openrouter.ai/api/v1

 |  
|

/nvd | NVIDIA NIM https://integrate.api.nvidia.com/v1

 |  
使用示例

bash  
# 通过代理调用百炼 API  
curl http://127.0.0.1:18888/bailian/chat/completions \  
  -H "Authorization: Bearer $BAILIAN_API_KEY" \  
  -H "Content-Type: application/json" \  
  -d '{"model": "qwen-plus", "messages": [{"role": "user", "content": "Hello"}]}'  
# 通过代理调用 OpenRouter  
curl http://127.0.0.1:18888/openrouter/chat/completions \  
  -H "Authorization: Bearer $OPENROUTER_API_KEY" \  
  -H "Content-Type: application/json" \  
  -d '{"model": "anthropic/claude-3-opus", "messages": [{"role": "user", "content": "Hello"}]}'

# 通过代理调用 NVIDIA NIM curl http://127.0.0.1:18888/nvd/chat/completions \ -H "Authorization: Bearer $NVIDIA_API_KEY" \ -H "Content-Type: application/json" \ -d '{"model": "meta/llama-3.1-8b-instruct", "messages": [{"role": "user", "content": "Hello"}]}'`

`健康检查`


HTTP 端点

bash  
# 健康状态  
curl http://127.0.0.1:18888/health  
# 统计数据  
curl http://127.0.0.1:18888/stats

  
健康检查响应

json  
{  
  "status": "ok",  
  "stats": {  
    "total_requests": 1234,  
    "total_responses": 1200,  
    "blocked": 5,  
    "warnings": 12,  
    "errors": 34  
  },  
  "uptime": 3600,  
  "rules_loaded": {  
    "layer1": 6,  
    "layer2": 3,  
    "whitelist": 2  
  }  
}

  
内容安全审计  
两层审核机制  
第一层：恶意指令检测  
危险系统命令（rm -rf /、mkfs、dd 等）  
提权操作（sudo su、chmod u+s）  
SQL 注入/删除  
数据外泄（curl -d @、nc -e）  
后门/反弹 shell  
第二层：敏感内容检测  
个人身份信息（身份证、手机号、邮箱）  
银行卡/信用卡号  
敏感关键词  
规则配置  
规则文件：

{baseDir}/scripts/content-filter-rules.json

json  
{  
  "layer1_malicious": {  
    "enabled": true,  
    "rules": [  
      {  
        "id": "CMD-001",  
        "name": "危险系统命令",  
        "severity": "critical",  
        "patterns": ["rm\\s+-rf\\s+[/~]", "mkfs\\.", "dd\\s+if=.of=/dev/"]  
      }  
    ]  
  },  
  "layer2_sensitive": {  
    "enabled": true,  
    "rules": [...]  
  },  
  "whitelist": ["已授权的.操作"]  
}

  
日志文件  
| 文件 | 内容 |  
|------|------|  
|

~/.openclaw/logs/llm-proxy/proxy-YYYY-MM-DD.jsonl

 | 请求日志 |  
|

~/.openclaw/logs/llm-proxy/service.log

 | 服务日志 |  
日志格式

json  
{  
  "timestamp": "2026-03-17T12:34:56.789",  
  "request_id": "abc12345",  
  "provider": "bailian",  
  "path": "/chat/completions",  
  "status": 200,  
  "duration_ms": 1234,  
  "alerts": [],  
  "request_size": 256,  
  "response_size": 1024  
}

  
常见问题  
端口被占用

bash  
# 查看占用端口的进程  
lsof -i :18888  
# 终止进程  
kill -9   
# 或重启代理（会自动清理）  
llm-proxy-ctl.sh restart

  
代理无响应

bash  
# 检查健康状态  
curl http://127.0.0.1:18888/health  
# 重启代理  
llm-proxy-ctl.sh restart

  
查看错误日志

bash  
tail -100 ~/.openclaw/logs/llm-proxy/service.log

  
核心文件  
| 文件 | 说明 |  
|------|------|  
|

{baseDir}/scripts/llm-proxy.py

 | 代理主程序 |  
|

{baseDir}/scripts/llm-proxy-ctl.sh

 | 启动控制脚本 |  
|

{baseDir}/scripts/content-filter-rules.json` | 内容过滤规则 |

--- 注意: 代理服务需要 Python 3.6+ 和网络连接。

数据来源：ClawHub ↗ · 中文优化：龙虾技能库