Tailor
Tailor Platform for AI assistants: search, list and read the platform documentation.
我该使用它吗
质量与安全性
发现（1）
- LOWTool 'fetch_knowledge' suggests web access but openWorldHint=false在 fetch_knowledge 中
基于对工具定义和协议合规性的自动分析。
上下文开销
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": {
"mcp": {
"url": "https://mcp.tailor.tech/"
}
}
}远程端点
https://mcp.tailor.tech/streamable-http它能做什么
工具清单
工具（3）
🟢fetch_knowledge(id, section)
Return the Markdown source of one documentation page, addressed by an id from search_knowledge or list_knowledge. The first line is the page's URL, for citation. Long pages come back as a heading outline instead; pass section to read one part of them.
输入模式
{
"type": "object",
"properties": {
"id": {
"type": "string",
"description": "page identifier exactly as returned by search_knowledge or list_knowledge, for example docs:sdk/quickstart.md"
},
"section": {
"type": "string",
"description": "return only this section of the page, matched case-insensitively against its level-two headings"
}
},
"required": [
"id"
],
"additionalProperties": false
}🟢list_knowledge(source, section)
List the documentation table of contents: the id, title, source, section and summary of every page. Use this to see what exists when a search does not find the right words; prefer search_knowledge when you know what you are looking for. Narrow the listing with source or section, since the full listing is large and is cut at 500 entries.
输入模式
{
"type": "object",
"properties": {
"source": {
"type": "string",
"description": "restrict the listing to one knowledge source; omit to list every source"
},
"section": {
"type": "string",
"description": "restrict the listing to one table-of-contents section, matched case-insensitively"
}
},
"additionalProperties": false
}🟢search_knowledge(query, source, limit)
Search the body text of Tailor Platform's public documentation and return matching pages with an excerpt. Start here for most questions, then pass an id to fetch_knowledge to read a page in full. Matching is literal and case-insensitive, so prefer the exact terms that would appear in the docs. This covers published documentation only; it cannot see a user's own application data.
输入模式
{
"type": "object",
"properties": {
"query": {
"type": "string",
"description": "words to look for in the documentation body, matched case-insensitively as a substring"
},
"source": {
"type": "string",
"description": "restrict the search to one knowledge source; omit to search every source"
},
"limit": {
"type": "integer",
"description": "maximum number of pages to return; defaults to 10, capped at 50"
}
},
"required": [
"query"
],
"additionalProperties": false
}社区
证据