A local, high-speed hybrid search engine designed for indexing and retrieving Markdown notes and documentation collections.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install qmd-skill-2
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 qmd-skill-2 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 qmd?
qmd is a specialized search engine built for developers and knowledge workers who manage extensive Markdown-based documentation. As a key component of the Openclaw Skills library, it enables users to perform lightning-fast keyword searches (BM25) and deep semantic vector searches on their local file systems. It bridges the gap between static files and active AI agents by providing a structured way to retrieve relevant context from personal knowledge bases.
The tool is privacy-centric, running entirely on your local hardware using GGUF models for embeddings and re-ranking. By indexing collections once, users can query their entire history of notes with minimal latency, making it an essential utility for anyone looking to augment their AI coding workflows with proprietary or personal documentation.
qmd Use Cases
- Searching personal notes, technical documentation, or long-form research stored in Markdown format.
- Finding semantically related content when specific keywords are forgotten or unknown.
- Retrieving full document contents to provide context for AI-driven coding or writing tasks.
- Automating the indexing of local knowledge bases to ensure search results are always up to date.
How qmd Works
- The user defines a collection by pointing qmd to a directory of Markdown files using a glob pattern.
- The engine performs content-based chunking, breaking documents into manageable segments optimized for LLM context windows.
- If semantic search is required, a local embedding process is run once to transform text into vector representations.
- When a search is triggered, qmd checks the index using BM25 for keyword matching or vector similarity for semantic intent.
- Results are returned with relevance scores, allowing the user or agent to fetch specific file contents by path or ID.
qmd Setup
Ensure you have Bun and SQLite installed on your system.
# Install qmd globally
bun install -g https://github.com/tobi/qmd
# Add your first collection
qmd collection add /path/to/notes --name notes --mask "**/*.md"
# (Optional) Generate embeddings for semantic search
qmd embed
qmd Data Schema & Taxonomy
qmd manages your data through collections and local caches. It organizes indexed content as follows:
| Component | Function |
|---|---|
| Collections | Named groups of files (e.g., 'work', 'journal') with specific file masks. |
| BM25 Index | Fast, keyword-based lookup table stored in a local SQLite database. |
| Vector Store | High-dimensional embeddings for semantic search capabilities. |
| Cache | Local GGUF models are stored in ~/.cache/qmd/models/. |
| JSON Output | Native support for machine-readable results via the --json flag. |
qmd Advanced Features
- Hybrid Querying: Leverage
qmd queryfor advanced reranking that combines keyword and semantic results for maximum precision. - Agent Integration: Utilize the
--jsonand--filesflags to integrate search results directly into other Openclaw Skills. - Automated Maintenance: Set up cron jobs for
qmd updateto keep your keyword index fresh without manual intervention. - Multi-Retrieval: Use
qmd multi-getwith glob patterns to pull content from multiple relevant files in a single command.
SKILL.md
Loading
Related Openclaw Skills
x-reader
A universal content processing engine that fetches, transcribes, and generates multi-dimensional insights from URLs, social media, and multimedia files.
Semantic Walk
A collaborative ritual for navigating semantic space to reach positions where complex insights become visible and accessible.
Flight Lines
A framework for navigating complex problems by composing deterritorialized operations from diverse domains through the Openclaw Skills ecosystem.
Akaunting
A comprehensive integration for managing bookkeeping, invoicing, and financial transactions through the Akaunting open-source platform.
QR Password
An air-gapped credential bridge that facilitates bidirectional secret transfer using QR codes as an optical channel.
agent-sims
A social simulation environment where AI agents with unique personalities interact, trade, and evolve within a shared virtual world.
Featured*
