Embedding Strategies

This one walks you through choosing and implementing embedding models for vector search, from OpenAI's text-embedding-3 variants to local options like BGE and sentence-transformers. It includes actual comparison tables showing dimensions and token limits, plus battle-tested chunking strategies: token-based, sentence-based, semantic sections, and recursive splitting. The code templates are production-ready, handling batching, normalization, and model-specific quirks like BGE's query prefixes. Useful if you're building RAG systems and need to move past "just use the default embeddings" into understanding tradeoffs between accuracy, cost, and speed. The chunking section alone will save you from reinventing overlap logic.

skills.sh installs: 210
GitHub stars: 39.4K
last updated: Jun 3, 2026

View on GitHub

Install

$ npx -y skills add sickn33/antigravity-awesome-skills --skill embedding-strategies --agent claude-code

Installs into .claude/skills of the current project.

Files

Embedding Strategies

Guide to selecting and optimizing embedding models for vector search applications.

Do not use this skill when

The task is unrelated to embedding strategies
You need a different domain or tool outside this scope

Instructions

Clarify goals, constraints, and required inputs.
Apply relevant best practices and validate outcomes.
Provide actionable steps and verification.
If detailed examples are required, open resources/implementation-playbook.md.

Use this skill when

Choosing embedding models for RAG
Optimizing chunking strategies
Fine-tuning embeddings for domains
Comparing embedding model performance
Reducing embedding dimensions
Handling multilingual content

Core Concepts

Model table is a historical candidate list, not a current leaderboard. Verify exact model-card token limits, dimensions, license, revision and query/document prefixes before use; count tokens with that model’s tokenizer. Approve private-document transfer/provider cost. Preserve original text and source offsets; stripping symbols can destroy code, monetary signs, equations and identifiers.