zettel-link
by hxy9243
Search & ResearchBrief
This skill maintains the Note Embeddings for Zettelkasten, to search notes, retrieve notes, and discover.
Description
What This Skill Does
Maintains semantic embeddings for an Obsidian vault to enable note search, retrieval, and connection discovery. Uses cosine similarity to find related notes across a Zettelkasten. Supports local (Ollama) and cloud (OpenAI, Gemini) embedding providers.
Semantic search finds conceptually related notes even when they share no keywords, which keyword search and manual tagging cannot do.
When to Use It
- Finding notes related to a half-remembered idea
- Discovering unexpected links between research notes
- Rebuilding broken connections after reorganizing a vault
- Searching notes by meaning rather than exact keywords
- Generating a graph of semantically connected notes
Example Workflow
User asks: find notes related to 'epistemology and belief formation'
- 1
Run config.py to confirm config/config.json exists
- 2
Run embed.py --input <vault_dir> to generate or update the embedding cache
- 3
Run search.py --input <vault_dir> --query 'epistemology and belief formation'
- 4
Read search results from .embeddings/search_results.json
- 5
Return the top matching note titles and snippets to the user
List of the 5 most semantically similar notes with similarity scores
Requirements
- Ollama installed locally with mxbai-embed-large model pulled (if using local provider)
- OPENAI_API_KEY environment variable (if using OpenAI provider)
- GEMINI_API_KEY environment variable (if using Gemini provider)
Installation
clawhub install hxy9243/zettel-linkRun the CLI command, or paste the prompt into Claude Code, Cursor, Codex or whichever coding agent you use.