PREFLIGHT — URL Routing for AI Agents
Choose HTTP, browser, machine endpoint, or avoid before an agent visits an unfamiliar URL.
我该使用它吗
质量与安全性
B
发现（1）
- LOWTool 'preflight_feedback' description lacks action verb在 preflight_feedback 中
基于对工具定义和协议合规性的自动分析。
上下文开销
~325token 数（工具定义）
~870 B典型响应大小
对注意力的影响极小（占 128k 上下文窗口的 0.25%）
这是每次将服务器的工具加载到模型上下文窗口时所消耗的大致 token 数。数值越高，可用于其他任务的注意力就越少。
安装
一键安装
将以下内容添加到你的 `claude_desktop_config.json` 文件中：
{
"mcpServers": {
"preflight": {
"url": "https://preflight-api-production-01a2.up.railway.app/mcp"
}
}
}远程端点
https://preflight-api-production-01a2.up.railway.app/mcpstreamable-http它能做什么
工具清单
工具（2）
🟢 只读🟡 写入🔴 删除⚪ 未知
🟢preflight_check(url)
Before fetching, crawling, scraping, opening, or browser-rendering an unfamiliar http/https URL, call this with the ACTUAL destination URL. Returns the best first route: HTTP, BROWSER, MACHINE_ENDPOINT, or AVOID, plus access/JS/size/cost hints. Do not substitute example.com when a real task URL is available.
输入模式
{
"type": "object",
"properties": {
"url": {
"type": "string",
"format": "uri",
"description": "Actual absolute http/https destination URL from the current task. Use before the first fetch or browser request."
}
},
"required": [
"url"
],
"additionalProperties": false
}⚪preflight_feedback(url, route, outcome, latencyMs)
After following a PREFLIGHT recommendation, report whether the chosen route worked so PREFLIGHT can improve future recommendations.
输入模式
{
"type": "object",
"properties": {
"url": {
"type": "string",
"format": "uri"
},
"route": {
"type": "string",
"enum": [
"HTTP",
"BROWSER",
"MACHINE_ENDPOINT",
"AVOID"
]
},
"outcome": {
"type": "string",
"enum": [
"success",
"failure"
]
},
"latencyMs": {
"type": "number",
"minimum": 0,
"maximum": 300000
}
},
"required": [
"url",
"route",
"outcome"
],
"additionalProperties": false
}社区
证据
最近观测
已验证未记录版本2 个工具
已验证未记录版本2 个工具