monitor
Check if an MCP server tool changed or hides injection patterns before you trust it.
我該用這個嗎
品質與安全性
根據工具定義與協定合規性的自動化分析。
上下文成本
這是每次將伺服器的工具載入模型上下文時所消耗的約略 token 數量。數量越高，可用於其他工作的注意力就越少。
安裝
一鍵安裝
將以下內容加入你的 `claude_desktop_config.json` 檔案：
{
"mcpServers": {
"monitor": {
"url": "https://tooldrift.agentexchange.work/mcp"
}
}
}遠端端點
https://tooldrift.agentexchange.work/mcpstreamable-http它能做什麼
工具清單
工具（3）
🟢check_mcp_server(url, server)
Check whether an MCP server's tools have changed since they were first recorded, and whether any tool description contains prompt-injection or data-exfiltration patterns. Call this BEFORE trusting a newly added or recently updated MCP server. Returns a verdict of clean or REVIEW_BEFORE_USE.
輸入結構描述
{
"type": "object",
"properties": {
"url": {
"type": "string",
"description": "The MCP server's HTTP endpoint to probe live."
},
"server": {
"type": "string",
"description": "Registry name of the server, if the URL is unknown."
}
}
}🟢get_drift_feed(limit)
List MCP tools across the public ecosystem whose definitions recently changed, newest first. Use to see whether a dependency mutated.
輸入結構描述
{
"type": "object",
"properties": {
"limit": {
"type": "integer",
"description": "Max rows, default 25."
}
}
}🟢get_findings(severity, limit)
List tools whose descriptions contain detected injection, concealment, or exfiltration patterns. Filter by severity: critical, high, medium, low.
輸入結構描述
{
"type": "object",
"properties": {
"severity": {
"type": "string"
},
"limit": {
"type": "integer"
}
}
}社群
證據