LOST

Local Offline Search Tool — a web search engine that runs entirely on your computer.

The database is stored locally, nothing is ever sent anywhere, and there's no search history. It keeps finding results with no internet connection at all.

Usage

Open desktop application (GUI)

lost

Running local web server

lost -s                             # web UI and Web API on http://127.0.0.1:1057
lost -s -a 0.0.0.0 -p 8080          # serve the local network on port 8080
lost -b                             # same as -s, but in the background
lost -w                             # run desktop app and web server together

Searching straight in the terminal

lost hacker news                    # Search for hacker news website
lost cache:https://slashdot.org     # View stored copy in the terminal
lost -f json rust -site:crates.io   # Output search results formatted as JSON
lost --suggest "rust comp"          # autocomplete
lost --suggest "how to _ catfish"   # autocomplete, fill in the blank

Working with datasets

lost -d hacker-news.csv           # import a BigQuery export of bigquery-public-data.hacker_news.full
lost -d CC-MAIN-…-00000.warc.wet.gz CC-MAIN-…-00000.warc.wat.gz   # import Common Crawl data
lost --imports                    # list past imports to see what's in the database
lost --remove-import 3            # remove import along with all pages that came from it
lost -e rust.lost rust compiler   # export what a search finds (or everything: lost -e all.lost)

Supported search operators:

• "exact phrase"
• "how to * catfish"
• -exclude
• OR
• ( )
• site:
• intitle:
• allintitle:
• inurl:
• allinurl:
• intext:
• allintext:
• filetype:
• link:

Data

LOST searches through whatever datasets you import. It comes out of the box with two pages (the front pages of Hacker News and Slashdot), so you could start seeing it in action right away.

Supported dataset formats

• Common Crawl (WET and WAT files)
• Hacker News (BigQuery CSV export)
• .lost files — LOST's own format

Datasets

Common Crawl

Common Crawl is a 501(c)(3) American non-profit organization that crawls the internet and provides a free, open-source archive of web data to the public. Their datasets are available for free to anyone here. The combined full size of their latest dataset is over 20TiB.

Here's one of the ways to obtain a small piece of their data:

curl -s https://data.commoncrawl.org/crawl-data/CC-MAIN-2026-39/wet.paths.gz | zcat | head -c 500000000 > sample.wet
curl -s https://data.commoncrawl.org/crawl-data/CC-MAIN-2026-39/wat.paths.gz | zcat | head -c 500000000 > sample.wat

Hacker News

The Hacker News public dataset is available on Google BigQuery. The dataset is part of the Google Cloud Marketplace Public Datasets and contains over 50 million rows (stories, comments, job posts) derived directly from the official Hacker News Firebase API. Merging this dataset into your LOST database will allow you to search through pages of Hacker News locally. The full size exceeds 50GiB.

Sample SQL command to select a chunk of data:

SELECT * FROM `bigquery-public-data.hacker_news.full` WHERE timestamp >= '2026-01-01' AND timestamp < '2026-10-10';

LOST files

LOST is capable of importing and exporting .lost dataset files, which makes it possible to freely and safely exchange crawl data with other LOST users. They can be optionally signed, to ensure that your database doesn't get corrupted or spammed. Both exports and imports can be made either via CLI (see Usage) or GUI (Settings → Data).

Database

The local database is stored by default either in ~/.lost/ (Linux, BSD), ~/Library/Application Support/LOST/ (macOS), or %USERPROFILE%\LOST\ (Windows).

Building

Everything

cargo build --release

Only CLI

cargo build --release --no-default-features --features cli

CLI & web server

cargo build --release --no-default-features --features cli,server

CLI & GUI

cargo build --release --no-default-features --features cli,gui

Bare library

cargo build --release --no-default-features --lib

Packages

make tarball     # (see the Makefile)
make deb         # needs cargo-deb
make appimage    # needs appimagetool

Installation

Download a package from the releases:

• Debian, Ubuntu, Mint: sudo apt install ./lost_…_amd64.deb
• Any Linux: the .AppImage — make it executable and run it. Or unpack the tarball and run sudo sh dist/install.sh.
• macOS: unzip, drag LOST to Applications. The first time, right-click it and choose Open (it isn't signed yet).
• Windows: unzip and run lost.exe.

To run the web server as a service (Linux, after installing the .deb or with install.sh): sudo systemctl enable --now lost, then open http://127.0.0.1:1057.

From source: make && sudo make install (sudo make uninstall to remove).

License

0BSD.

res/web/mikado.light.min.js is Mikado, © Nextapps GmbH, Apache-2.0.