Dr. QuantMaster MCP Server

import { getCollection, isVectorSearchAvailable } from "./client.js"; import { COLLECTIONS, CollectionCategory } from "./collections.js"; import type { Metadata } from "chromadb"; export interface SearchResult { content: string; metadata: Record<string, string | number | boolean>; distance?: number; } export async function searchKnowledgeBase( query: string, category: CollectionCategory | "all" = "all", nResults: number = 5 ): Promise<SearchResult[]> { // If ChromaDB is not available, return empty results with a note if (!isVectorSearchAvailable()) { return [{ content: "Vector search is currently unavailable. The tool is operating without RAG support. To enable, start ChromaDB server: chroma run --path ./chroma-data", metadata: { source: "system", type: "notice" }, distance: 0 }]; } const collectionsToSearch = category === "all" ? Object.values(COLLECTIONS).map((c) => c.name) : Object.values(COLLECTIONS) .filter((c) => c.metadata.category === category) .map((c) => c.name); const results: SearchResult[] = []; for (const collectionName of collectionsToSearch) { try { const collection = await getCollection(collectionName); if (!collection) continue; const queryResult = await collection.query({ queryTexts: [query], nResults: nResults, }); if (queryResult.documents[0]) { results.push( ...queryResult.documents[0].map((doc, i) => ({ content: doc || "", metadata: (queryResult.metadatas[0]?.[i] || {}) as Record<string, string | number | boolean>, distance: queryResult.distances?.[0]?.[i], })) ); } } catch (error) { // Silently skip failed collections } } // Sort by relevance (lower distance = more relevant) results.sort((a, b) => (a.distance || 0) - (b.distance || 0)); return results.slice(0, nResults); } export async function addDocument( collectionName: string, id: string, content: string, metadata: Metadata ): Promise<void> { if (!isVectorSearchAvailable()) return; const collection = await getCollection(collectionName); if (!collection) return; await collection.add({ ids: [id], documents: [content], metadatas: [metadata], }); } export async function addDocuments( collectionName: string, ids: string[], contents: string[], metadatas: Metadata[] ): Promise<void> { if (!isVectorSearchAvailable()) return; const collection = await getCollection(collectionName); if (!collection) return; await collection.add({ ids, documents: contents, metadatas, }); }