The Engineer's Guide to Local Documentation Search: Getting Started with tobi/qmd
Here is a breakdown of why this tool is a gem for engineers and how to get it running.
In the engineering world, we usually choose between two extremes
a basic grep (which is fast but "dumb") or a heavy enterprise search tool (which is "smart" but slow and invades privacy).
tobi/qmd hits the sweet spot
Local-First (Privacy)
Your internal system designs and sensitive meeting notes never leave your machine. No API keys or cloud leakage.
SOTA (State of the Art) Performance
It uses modern retrieval techniques (like vector embeddings or advanced indexing) to understand context, not just keywords.
CLI-Centric
You don't have to leave your terminal or open a bulky browser tab. It stays right in your workflow.
Since it's a CLI tool, the setup is usually straightforward. Assuming you have a Go or Python environment (depending on the specific build), you can typically install it via a package manager or by cloning the repo.
Clone the Repository
git clone https://github.com/tobi/qmd.git
cd qmd
Build/Install
(Check the specific README in the repo for the latest build command, but usually it's
)
make build
# or
go install ./...
The magic happens when you point qmd at your "Second Brain" folder.
First, you need to let the engine "read" your files.
qmd index ~/Documents/my-notes/ --include "*.md"
Instead of remembering the exact filename, you can ask a question
qmd search "How does the auth microservice handle JWT rotation?"
As engineers, we love automation. You can add an alias to your .zshrc or .bashrc so you can search your docs from any directory instantly.
# Add this to your shell config
alias ?='qmd search'
# Now, in your terminal, just type:
? "deployment steps for staging"
If you were to look at how a tool like this works, it likely follows this architectural flow
Parsing
It reads your Markdown or text files.
Embedding
It converts text into mathematical vectors that represent meaning.
Vector Store
It saves these locally.
Querying
When you search, it compares your query's vector to your notes' vectors and returns the closest matches.
qmd is basically like having a personal Google for your own local files. It’s fast, private, and stays out of your way until you need it.