1.4.0

Author: binggg

mcp/package-lock.json

@@ -1,6 +1,6 @@
 {
   "name": "@cloudbase/cloudbase-mcp",
-  "version": "1.3.0",
+  "version": "1.4.0",
   "description": "腾讯云开发 MCP Server，支持静态托管/环境查询/",
   "main": "index.js",
   "type": "module",

mcp/package.json

@@ -0,0 +1,124 @@
+import { z } from "zod";
+import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
+
+// 1. 枚举定义
+const KnowledgeBaseEnum = z.enum(["cloudbase", "scf"]);
+// 2. 枚举到后端 id 的映射
+const KnowledgeBaseIdMap: Record<z.infer<typeof KnowledgeBaseEnum>, string> = {
+  cloudbase: "ykfzskv4_ad28",
+  scf: "scfsczskzyws_4bdc",
+};
+
+// 安全 JSON.parse
+function safeParse(str: string) {
+  try {
+    return JSON.parse(str);
+  } catch (e) {
+    return {};
+  }
+}
+
+// 安全 JSON.stringify，处理循环引用
+function safeStringify(obj: any) {
+  const seen = new WeakSet();
+  try {
+    return JSON.stringify(obj, function(key, value) {
+      if (typeof value === "object" && value !== null) {
+        if (seen.has(value)) return;
+        seen.add(value);
+      }
+      return value;
+    });
+  } catch (e) {
+    return "";
+  }
+}
+
+export function registerRagTools(server: McpServer) {
+    // 知识库检索
+    server.tool(
+        'search_knowledge_base', 
+        '云开发知识库智能检索工具，支持云开发与云函数知识的向量查询',
+    {
+        threshold: z.number().default(0.5).optional().describe("相似性检索阈值"),
+        id: KnowledgeBaseEnum.describe("知识库范围，cloudbase=云开发全量知识，scf=云开发的云函数知识"),
+        content: z.string().describe("检索内容"),
+        options: z.object({
+            chunkExpand: z.array(z.number()).min(2).max(2).default([3, 3]).describe("指定返回的文档内容的展开长度,例如 [3,3]代表前后展开长度"),
+        }).optional().describe("其他选项"),
+        limit: z.number().default(5).optional().describe("指定返回最相似的 Top K 的 K 的值")
+    }, 
+    async ({
+        id,
+        content,
+        options: {
+            chunkExpand = [3, 3]
+        } = {},
+        limit = 5,
+        threshold = 0.5
+    }) => {
+        // 枚举到后端 id 映射
+        const backendId = KnowledgeBaseIdMap[id as keyof typeof KnowledgeBaseIdMap] || id;
+        const signInRes = await fetch('https://tcb-advanced-a656fc.api.tcloudbasegateway.com/auth/v1/signin/anonymously',{
+            method: 'POST',
+            headers: {
+                'Content-Type': 'application/json',
+                'Accept': 'application/json',
+                'x-device-id': 'cloudbase-ai-toolkit'
+            },
+            body: safeStringify({
+                collectionView: backendId,
+                options: {
+                    chunkExpand
+                },
+                search: {
+                    content: content,
+                    limit
+                }
+            })
+        })
+        const token = (await signInRes.json()).access_token
+        const res = await fetch(`https://tcb-advanced-a656fc.api.tcloudbasegateway.com/v1/knowledge/search`, {
+            method: 'POST',
+            headers: {
+                'Content-Type': 'application/json',
+                'Authorization': `Bearer ${token}`
+            },
+            body: safeStringify({
+                collectionView: backendId,
+                options: {
+                    chunkExpand
+                },
+                search: {
+                    content: content,
+                    limit
+                }
+            })
+        })
+        const result = await res.json()
+
+        if (result.code) {
+            throw new Error(result.message)
+        }
+        console.log(result)
+        return {
+            content: [{
+                type: "text",
+                text: safeStringify(result.data.documents
+                    .filter((item: any) => item.score >= threshold)
+                    .map((item: any) => {
+                        return {
+                            score: item.score,
+                            fileTile: item.documentSet.fileTitle,
+                            url: safeParse(item.documentSet.fileMetaData).url,
+                            paragraphTitle: item.data.paragraphTitle,
+                            text:`${item.data.pre?.join('\n') || ''}
+${item.data.text}
+${item.data.next?.join('\n') || ''}`,
+                        }
+                    })
+                )
+            }]
+        }
+    })
+}