创思大模型安全 MCP
A content security protection system for large language models, providing real-time risk identification and interception to ensure safe, compliant, and trustworthy model applications.
创思大模型安全 MCP
官网
简介
创思大模型安全 MCP(Model Content Protection)是面向大语言模型的内容安全防护系统,通过实时风险识别与拦截,保障大模型应用的安全、合规与可信。
核心能力:
- 输入/输出内容双重安全检测
- 多维度风险识别(合规/伦理/安全)
- 低延迟实时防护
- 可定制策略引擎
快速接入
在配置文件中添加以下服务配置(支持 SSE 协议):
{
"mcpServers": {
"chuangsiai": {
"name": "chuangsiai-mcp",
"type": "sse",
"url": "https://mcp.chuangsiai.com/sse",
"headers": {
"Authorization": "【你的API Key】",
"StrategyKey": "【你的策略标识】"
}
}
}
}
凭证获取步骤
技术特性
- 实时风险识别:通过实时分析输入和输出内容,快速识别潜在的风险,如违规内容、伦理问题等。
- 多维度检测:支持多种风险检测维度,包括但不限于合规性、伦理性和安全性。
- 低延迟:采用 HTTPS 传输协议,确保低延迟的实时防护,满足实时性要求。
- 可定制策略引擎:提供灵活的策略引擎,支持用户自定义风险检测规则和防护策略。
使用场景
- 内容审核:实时监控和审核输入和输出内容,确保内容符合法律法规和社会主义核心价值观。
📬 联系我们
如需技术支持、企业合作或 API 接入,请联系:
Serveurs connexes
ESA MCP Server
An MCP server for Alibaba Cloud's Edge Security Acceleration (ESA) service.
DeAI API MCP Server
Interact with the DeAI API for token analysis, holder tracking, and wallet portfolio analytics on Ethereum.
notebooklm-mcp-secure
Security-hardened NotebookLM MCP with post-quantum encryption, GDPR/SOC2 compliance, and 14 security layers. Query Google's Gemini-grounded research from any MCP-compatible AI assistant.
BICScan
Provides risk scoring and asset information for blockchain addresses, domains, and dApps using the BICScan API.
CData LinkedIn Ads
MCP Server for LinkedIn Ads, powered by the CData JDBC Driver. Requires a separate license and configuration.
Linode MCP Server
Manage Linode cloud infrastructure resources through natural language conversation.
Remote MCP Server (Authless)
An example of a remote MCP server deployable on Cloudflare Workers without authentication.
AWS MCP
A natural language interface to manage AWS services like RDS, S3, EC2, and VPC.
Keycloak MCP Server
Administer Keycloak users and realms using a Model Context Protocol server.
Prismic
An MCP server for interacting with the Prismic headless CMS.