voyage-embeddings
Installation
SKILL.md
Voyage AI embeddings
Voyage AI is the embedding provider Anthropic recommends as a complement to Claude. This skill covers two things:
- Reference — which model to pick, the
input_typerule, dimensions, quantization - Workflow — embedding text/code with Voyage and upserting into a standard (non-integrated) Pinecone index
When this skill applies
- User wants higher retrieval quality than
llama-text-embed-v2provides — especially for code (voyage-code-3), finance (voyage-finance-2), or legal (voyage-law-2) content - User is embedding content client-side and pushing vectors to Pinecone
- User is migrating an existing integrated index to a Voyage model
- User asks any "which model" or "what dimensions" question about Voyage
When this skill does NOT apply
- User wants to query an integrated Pinecone index with a built-in embedder (use the
pinecone-queryskill — that one uses MCP) - User wants generic OpenAI / Cohere / HuggingFace embeddings — different provider, different SDK