data-ai MCP Server
AI Agent Security Middleware — 8-layer defense, DLP data flow, prompt injection detection, zero dependencies. SDK + MCP server for Claude Code, Cursor, LangChain, Hermes Agent & more.
Discovered via list:awesome-mcp-servers-punkpeye and last synced 3mo ago.
1. Install the package
npx shellward
2. Add to claude_desktop_config.json
{
"mcpServers": {
"shellward": {
"command": "npx",
"args": [
"shellward"
]
}
}
}Config file location: ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) / %APPDATA%\Claude\claude_desktop_config.json (Windows)
Integration
Check if file path operation is safe (.env, .ssh, credentials...)
`auto` / `zh` / `en`
Description
`auto` / `zh` / `en`
ShellWard
`npm install shellward`，3 行代码接入
Check if tool name is allowed (blocks payment/transfer tools)
`0`-`100`
Detect prompt injection in text (37+ rules, zh+en)
Values
3
`enforce` / `audit`
183 passing (incl. 15 MCP + 12 ReDoS + live tool-poisoning scan)
[agentguard](https://github.com/GoPlusSecurity/agentguard)
Values
0
Check if a shell command is safe (rm -rf, reverse shell, fork bomb...)
`0`-`100`
❌
Scan for PII & sensitive data (CN ID/phone/bank, API keys, SSN...)
`enforce` / `audit`
24
Audit AI response for canary leaks & PII exposure
Description
Get current security config & active layers
Data
Precision
24
Scan an MCP tool definition for poisoning + rug-pull
file path