Does this support different embedding models?

Yes, the skill is designed to handle model selection and configuration to ensure the correct task types are applied regardless of the underlying provider.

What is an asymmetric embedding strategy?

It is a technique where different task types are applied to documents and queries to ensure the mathematical vector space is optimized for retrieving long-form content using short-form questions.

How does this skill improve RAG accuracy?

By enforcing RETRIEVAL_DOCUMENT for ingestion and RETRIEVAL_QUERY for searches, it ensures that Claude's retrieval logic aligns with the specific intent of the data being processed.

Why shouldn't I use Batch APIs for document embeddings in RAG?

Many Batch APIs default to query-based task types, which causes a semantic mismatch when used for document indexing, leading to poor retrieval results.

Embedding Strategy Optimization

Name: Embedding Strategy Optimization
Author: Agentient

byAgentient

Ciencia de Datos y ML

Implements asymmetric embedding strategies to improve vector search accuracy by distinguishing between document ingestion and user queries.

Acerca de

This skill optimizes RAG (Retrieval-Augmented Generation) pipelines by enforcing asymmetric embedding patterns, ensuring that documents are ingested with specific document task types while queries use dedicated query task types. By preventing the common pitfall of using batch API defaults or symmetric embedding for both indexing and retrieval, it eliminates semantic mismatches and significantly improves the relevance and precision of retrieved information in AI-driven applications.

Características Principales

Optimized document ingestion via concurrent online embedding APIs
Automated embedding model selection and configuration
Validation of task type alignment to ensure optimal retrieval quality
Asymmetric task type enforcement for document vs. query embeddings
Semantic mismatch prevention for high-precision RAG pipelines
0 GitHub stars

Casos de Uso

Optimizing RAG performance in enterprise-grade AI applications
Building high-precision vector search for technical documentation or knowledge bases
Correcting semantic drift caused by improper batch embedding defaults

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add agentient/vibekit embedding-strategy

For use in Claude.ai and ChatGPT

Download Skill

GitHub