pixserp
Live AI-native web search with citations. One tool for every MCP client. Flat per-request pricing.
我该使用它吗
质量与安全性
A
基于对工具定义和协议合规性的自动分析。
上下文开销
~265token 数（工具定义）
~2.1 KB典型响应大小
对注意力的影响极小（占 128k 上下文窗口的 0.21%）
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": {
"search": {
"url": "https://pixserp.com/api/v1/mcp"
}
}
}远程端点
https://pixserp.com/api/v1/mcpstreamable-http它能做什么
工具清单
工具（1）
🟢 只读🟡 写入🔴 删除⚪ 未知
🟢search(query, model, max_steps)
Search the live web and get an AI-synthesized answer with structured citations. One endpoint covers 10 answer shapes — web, news, images, places & maps, shopping, flights, hotels, YouTube, transcripts, any URL — the right one is picked automatically from your query. Use this whenever you need fresh, factual information from the web.
输入模式
{
"type": "object",
"properties": {
"query": {
"type": "string",
"description": "What to search for. Phrase as a natural-language question or a specific topic."
},
"model": {
"type": "string",
"enum": [
"pixserp-fast",
"pixserp-standard",
"pixserp-deep",
"pixserp-agent"
],
"default": "pixserp-fast",
"description": "`pixserp-fast` = quick single-search lookup (default). `pixserp-standard` = balanced. `pixserp-deep` = multi-angle thorough research. `pixserp-agent` = multi-step research loop, decides itself when to stop."
},
"max_steps": {
"type": "integer",
"minimum": 1,
"maximum": 100,
"description": "Only used when `model = pixserp-agent`. Caps the research loop length. Default 50, max 100."
}
},
"required": [
"query"
],
"additionalProperties": false
}社区
证据
最近观测
已验证未记录版本1 个工具
已验证未记录版本1 个工具
已验证未记录版本1 个工具