Vector Memory

This gives you semantic search over stored text without setting up a vector database. It generates embeddings via OpenAI's API and stores them in JSON, then searches using cosine similarity to find memories by meaning rather than exact keywords. Useful if you're building an agent that needs to recall relevant context from past conversations or documents. The trade-off is obvious: you get simplicity and zero infrastructure, but you're limited to in-memory search that won't scale past a few thousand entries. For prototyping or moderate memory needs, though, it's a clean way to add semantic recall without the database overhead.

Install to Claude Code

npx -y skills add winsorllc/upgraded-carnival --skill vector-memory --agent claude-code