A semantic search and connection discovery tool for Obsidian vaults using local or cloud-based AI embeddings.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install zettel-link
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install zettel-link using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Zettel Link?
Zettel Link is a specialized tool designed for knowledge workers and researchers who utilize the Zettelkasten method for personal knowledge management. It leverages advanced vector embedding models to transform static Markdown notes into a semantically searchable network, allowing for conceptual retrieval that goes beyond traditional keyword matching. As a powerful addition to the Openclaw Skills ecosystem, it provides the necessary scripts to maintain an intelligent cache of your notes, making it easier to navigate complex information architectures.
The skill supports a variety of embedding providers, ranging from local-first solutions like Ollama to industry-standard APIs from OpenAI and Google Gemini. By automating the process of embedding and calculating cosine similarity between files, Zettel Link helps users uncover hidden connections and surface relevant content exactly when it is needed most in their research workflow.
Zettel Link Use Cases
- Performing semantic searches across an Obsidian vault when specific keywords are difficult to recall.
- Identifying hidden conceptual links between permanent notes to strengthen a Zettelkasten network.
- Generating automated suggestions for related notes based on thematic similarity.
- Maintaining a high-performance local embedding cache for offline AI-assisted research.
- Auditing a digital garden to find orphaned notes that share high similarity with existing clusters.
How Zettel Link Works
- The user initializes the skill by running the configuration script to define the preferred embedding provider and model parameters.
- The embedding script scans the target directory, processing new or modified Markdown files while respecting exclusion rules for system folders.
- Vector representations are generated for each note and stored in a local JSON cache, ensuring subsequent operations are fast and cost-effective.
- When a search is triggered, the query is embedded and compared against the cache using cosine similarity to find the most relevant matches.
- The linking script performs a pair-wise similarity analysis across the entire vault to discover connections that exceed a user-defined threshold.
- Results are output to structured JSON files within the vault, allowing other tools in the Openclaw Skills suite to consume the data.
Zettel Link Setup
To integrate this tool into your Openclaw Skills workflow, ensure you have Python 3.10+ and the uv package manager installed.
First, initialize your configuration:
uv run scripts/config.py
If you prefer to use a specific provider like OpenAI or Gemini, configure it with:
uv run scripts/config.py --provider openai --model text-embedding-3-small
Generate your initial embeddings by pointing the script to your notes directory:
uv run scripts/embed.py --input /path/to/your/vault
Zettel Link Data Schema & Taxonomy
The skill manages its metadata and vector storage within a hidden .embeddings/ directory located at the root of your target folder. This ensures that the core Openclaw Skills data remains portable and localized.
| File | Purpose |
|---|---|
embeddings.json |
The primary cache containing vector arrays, file metadata, and mtime timestamps for incremental updates. |
links.json |
A compiled list of semantic connections between notes, grouped by file and sorted by similarity score. |
search_results.json |
Stores the output of the most recent semantic search operation for agent retrieval. |
config/config.json |
Global settings for the skill, including model selection, top_k values, and skip directories. |
Zettel Link Advanced Features
- Incremental update logic that only re-embeds files modified since the last run, significantly reducing processing time.
- Intelligent text truncation that automatically handles notes exceeding the maximum token limits of specific embedding models.
- Adjustable similarity thresholds to fine-tune the density of discovered connections within a Zettelkasten.
- Support for multiple local and remote providers, allowing users to switch between cost-effective local models and high-accuracy cloud APIs.
- Idempotent script design, making it safe for agents within the Openclaw Skills framework to re-run processes without data loss.
SKILL.md
Loading
Related Openclaw Skills
Weibo Timeline Monitor
Automates the monitoring and archiving of the Weibo Latest Posts timeline into organized, deduplicated Markdown files.
Claude Code Integration
A comprehensive integration for OpenClaw that brings Claude Code's advanced AI-assisted coding, documentation queries, and subagent management into your workflow.
Xiaohongshu Video Downloader
An automated utility for extracting and downloading high-quality videos from Xiaohongshu (Red) using Openclaw Skills.
Skills Firewall
An automated security layer designed to scan, filter, and block malicious or harmful skills before they compromise your system.
Model Matrix
A weighted, policy-aware routing system for selecting the optimal AI model based on real-world performance, cost, and community sentiment.
Oblien Workspace Runtime
A secure, isolated Firecracker microVM environment providing a robust Internal API for AI agents to manage files, execute commands, and run interactive terminals.
METADATA
Featured*
