KS
Killer-Skills

managing-embeddings — how to use managing-embeddings how to use managing-embeddings, what is managing-embeddings, managing-embeddings alternative, managing-embeddings vs OpenAI, managing-embeddings install, managing-embeddings setup guide, Gemini embedding for AI agents, Anki card management with embeddings, semantic similarity tasks with managing-embeddings

v1.0.0
GitHub

About this Skill

Perfect for AI Agents needing advanced embedding management capabilities with Gemini embedding Managing-embeddings is a technical skill for dividing Anki cards into atomic units using Gemini embedding, specifically designed for AI agents.

Features

Utilizes the Gemini embedding model (gemini-embedding-001) for semantic similarity tasks
Operates with a 768-dimensional embedding space (default value)
Accepts input limits of up to 8K tokens
Supports taskType specification as SEMANTIC_SIMILARITY
Directly calls the @google/genai library (GoogleGenAI) for embedding management
Features a modular structure with client.ts and cosine.ts files for Gemini embedding API client and cosine similarity calculations

# Core Topics

shren207 shren207
[0]
[0]
Updated: 3/7/2026

Quality Score

Top 5%
51
Excellent
Based on code quality & docs
Installation
SYS Universal Install (Auto-Detect)
Cursor IDE Windsurf IDE VS Code IDE
> npx killer-skills add shren207/awesome-anki/managing-embeddings

Agent Capability Analysis

The managing-embeddings MCP Server by shren207 is an open-source Categories.community integration for Claude and other AI agents, enabling seamless task automation and capability expansion. Optimized for how to use managing-embeddings, what is managing-embeddings, managing-embeddings alternative.

Ideal Agent Persona

Perfect for AI Agents needing advanced embedding management capabilities with Gemini embedding

Core Value

Empowers agents to manage embeddings using Gemini embedding with 768 dimensions and semantic similarity task type, directly calling GoogleGenAI for specialized embedding management, allowing for efficient division of Anki cards into atomic units

Capabilities Granted for managing-embeddings MCP Server

Dividing Anki cards into atomic units using Gemini embedding
Managing semantic similarity tasks with 8K token input limit
Optimizing embedding dimensions for specific AI tasks

! Prerequisites & Limits

  • Limited to Gemini embedding, cannot be switched to other providers like OpenAI
  • Requires direct call to GoogleGenAI, bypassing LLM abstraction layer
Project
SKILL.md
2.6 KB
.cursorrules
1.2 KB
package.json
240 B
Ready
UTF-8

# Tags

[No tags]
SKILL.md
Readonly

임베딩 관리

기술 스택

  • 모델: gemini-embedding-001 (GA, MTEB 상위권)
  • 차원: 768 (기본값)
  • 입력 한도: 8K 토큰
  • taskType: SEMANTIC_SIMILARITY (문자열로 직접 지정)

LLM 추상화 미사용

Note: 임베딩은 packages/core/src/llm/ 추상화 계층을 사용하지 않고, packages/core/src/embedding/client.ts에서 @google/genai (GoogleGenAI)를 직접 호출합니다. 따라서 임베딩은 Gemini 전용이며, OpenAI 등 다른 프로바이더로 전환할 수 없습니다.

모듈 구조 (packages/core/src/embedding/)

파일역할
client.tsGemini 임베딩 API 클라이언트
cosine.ts코사인 유사도 계산 (0-100%)
cache.ts파일 기반 증분 캐시

주요 함수

typescript
1// 단일 텍스트 임베딩 2const embedding = await getEmbedding(text); // number[] (768차원) 3 4// 의미적 유사도 5const similarity = await getSemanticSimilarity(text1, text2); // 0-100 (%) 6 7// 유사성 검사 (임베딩 모드) 8const result = await checkSimilarity(targetCard, allCards, { 9 useEmbedding: true, deckName: '덱명', threshold: 85 10});

텍스트 전처리

임베딩 생성 전 반드시 정리:

  1. Cloze 구문에서 내용만 추출 ({{c1::DNS}}DNS)
  2. HTML 태그 제거
  3. 컨테이너 구문 제거 (::: tip 등)
  4. nid 링크에서 제목만 추출

캐시 전략

  • 저장 위치: output/embeddings/{deckNameHash}.json
  • 구조: { [noteId]: { embedding, textHash, timestamp } }
  • 증분 업데이트: MD5 해시로 텍스트 변경 감지, 변경된 카드만 재생성
  • 캐시 확인: GET /api/embedding/status/:deckName
  • 캐시 삭제: DELETE /api/embedding/cache/:deckName

API 엔드포인트

MethodPath설명
POST/api/embedding/generate덱 전체 임베딩 생성 (증분)
GET/api/embedding/status/:deckName캐시 상태 확인
DELETE/api/embedding/cache/:deckName캐시 삭제
POST/api/embedding/single단일 텍스트 임베딩 (디버깅)

자주 발생하는 문제

  • TaskType 미지원: @google/genai에서 TaskType enum이 export 안 됨 → 문자열 'SEMANTIC_SIMILARITY'로 직접 지정
  • 캐시 위치 혼동: 덱 이름을 MD5 해시로 변환하여 파일명 생성

상세 참조

  • references/embedding-system.md — gemini-embedding-001, 캐시 전략 상세
  • references/preprocessing.md — Cloze/HTML/컨테이너 제거 로직
  • references/troubleshooting.md — TaskType 미지원, 캐시 위치

Related Skills

Looking for an alternative to managing-embeddings or building a Categories.community AI Agent? Explore these related open-source MCP Servers.

View All

widget-generator

Logo of f
f

widget-generator is an open-source AI agent skill for creating widget plugins that are injected into prompt feeds on prompts.chat. It supports two rendering modes: standard prompt widgets using default PromptCard styling and custom render widgets built as full React components.

149.6k
0
Design

chat-sdk

Logo of lobehub
lobehub

chat-sdk is a unified TypeScript SDK for building chat bots across multiple platforms, providing a single interface for deploying bot logic.

73.0k
0
Communication

zustand

Logo of lobehub
lobehub

The ultimate space for work and life — to find, build, and collaborate with agent teammates that grow with you. We are taking agent harness to the next level — enabling multi-agent collaboration, effortless agent team design, and introducing agents as the unit of work interaction.

72.8k
0
Communication

data-fetching

Logo of lobehub
lobehub

The ultimate space for work and life — to find, build, and collaborate with agent teammates that grow with you. We are taking agent harness to the next level — enabling multi-agent collaboration, effortless agent team design, and introducing agents as the unit of work interaction.

72.8k
0
Communication