SearchHub

A local search engine for your browser bookmarks and history. Import bookmarks/history from Firefox, Zen, Chrome, or Chromium, search them with full-text queries, and optionally forward searches to external engines like wikipedia or SearXNG (aggregates results from dozens of backends). Content can be automatically tagged via local ONNX embeddings (opt-in; set tagging_enabled = true in config).

The ONNX tagging model is embedded via the optional tagging feature (included by default). If you don't need auto-tagging, you can build a ~30% smaller binary by excluding it.

Install

Binaries are available at vit.am/~ololduck/search_hub/latest. Download the binary for your architecture, extract, and run.

Source: Clone the repository and build with Rust:

Prerequisites: Rust (install via rustup).

git clone https://vit.am/~ololduck/search_hub/repository.git search_hub
cd search_hub
cargo install --path .

This installs the search_hub binary to ~/.cargo/bin/search_hub.

Smaller binary (no ONNX tagging): If you don't need auto-tagging, skip the ~12 MiB ONNX Runtime dependency:

cargo install --path . --no-default-features

The resulting binary still supports all features except semantic tagging. Any search_hub retag or auto-tagging operation will advise you to rebuild with --features tagging.

To update later, pull the latest code and reinstall.

First steps

# Import bookmarks from Firefox (auto-discovers your profile)
search_hub import firefox

# Import from Chrome
search_hub import chrome

# Start the web UI
search_hub serve

Open http://127.0.0.1:8080 in your browser. You can now search your bookmarks.

Search queries are also forwarded to external engines: Wikipedia, crates.io via its public JSON API, and optionally SearXNG (which aggregates Google, Bing, DDG, and dozens more) if [[engines]] is configured. Works as a custom search provider in Firefox/Zen via the OpenSearch protocol (your browser should auto-discover it at /opensearch.xml).

When JavaScript is available, search results are progressively loaded: local bookmark results appear instantly, then external engine results stream in as each engine responds.

The server supports a ?js=true query parameter: when present, it skips external engines during page render (returning only local results in ~50ms) and lets the client fetch them via a streaming connection. The inline JavaScript automatically adds ?js=true to the URL, so the fast path is active by default once JS runs.

For the fastest initial page load from your browser's search bar (before any JS runs), configure your browser's search engine URL with ?js=true appended:

• Firefox: Settings → Search → Manage Search Engines → Add, use http://localhost:8080/search?q=%s&js=true
• OpenSearch template: http://localhost:8080/search?q={searchTerms}&js=true

Without ?js=true, the server waits for all external engines before sending the page — a fallback path for browsers without JavaScript (which still works, just slower on first load).

CLI reference

All commands use ~/.local/share/search_hub/bookmarks.db by default. Override with --db-path or set db_path in the config file.

The first time you use a tagging-enabled command (search_hub insert, search_hub retag, or import with tagging_enabled = true), SearchHub downloads an ONNX embedding model to $XDG_CACHE_DIR (defaults to ~/.cache/search_hub) (about 127 MB). This only applies if the binary was built with the tagging feature (default: on).

Configuration

Run search_hub init-config to create ~/.config/search_hub/config.toml with all available options commented out. Or create it manually:

# Bookmark database path (default: platform data directory)
# db_path = "/home/you/.local/share/search_hub/bookmarks.db"

# Custom tags override the built-in defaults
# [[tags]]
# name = "my-custom-tag"
# examples = ["example text one", "example text two"]

# Whether auto-tagging is enabled (default: false, requires ONNX model download on first use)
# tagging_enabled = true

# Minimum confidence for auto-tagging (0.0 to 1.0, default: 0.6)
# tagging_threshold = 0.6

# Hosts to skip when fetching content for bookmarking (default: local addresses)
# exclude_urls = ["localhost", "127.0.0.1", "::1"]

# Per-engine configuration (optional)
# Multiple instances supported (e.g., public + private crates.io registries)
[[engines]]
type = "searxng"
instance = "https://search.kael.ink"
# timeout_secs = 10.0  # optional per-engine timeout
# Best: use an existing public instance (see https://searx.space).
# Also possible: run your own with Docker:
#   docker run -d --name searxng -p 8888:8080 searxng/searxng

# Custom crates.io registry (optional)
# [[engines]]
# type = "crates_io"
# url = "https://registry.example.com/api/v1/crates?q={}&per_page=10"
# timeout_secs = 5.0

# Wikipedia search (optional, defaults to English)
# [[engines]]
# type = "wikipedia"
# lang = "fr"
# timeout_secs = 5.0

# MDN Web Docs search (optional, defaults to en-US)
# [[engines]]
# type = "mdn"
# locale = "fr"
# timeout_secs = 5.0

# Generic HTML-scraped engine (use with any search site)
# Provide a URL template with `{}` for the query and a CSS selector
# targeting the result container. Results are extracted from `<a>` links
# inside that container (deduplicated, up to 10, http/https only).
#
# Note: most commercial search engines (Google, Bing, DuckDuckGo, etc.)
# block automated requests. This engine works best with small/niche sites
# that don't enforce bot detection. To find the right selector, view the
# page source or use browser dev tools on the search results page.
# [[engines]]
# type = "generic"
# name = "DuckDuckGo"
# url = "https://html.duckduckgo.com/html/?q={}"
# selector = "div.results"
# timeout_secs = 10.0
# shortcode = "ddg"           # optional: override auto-generated shortcode
# bang_enabled = true          # optional: disable ! redirect but keep @
# bang_url = "..."             # optional: custom redirect URL (keeps shortcode)

Search shortcuts

SearchHub supports two query prefixes that use shortcodes — compact aliases auto-generated from your configured [[engines]].

Auto-generated shortcodes

Each engine type gets a sensible default shortcode:

Overriding shortcodes per engine

Set shortcode, bang_url, or bang_enabled directly on the engine:

[[engines]]
type = "wikipedia"
lang = "fr"
shortcode = "wikifr"       # overrides "wfr"
bang_enabled = false       # disable ! redirect (still searchable via @)
bang_url = "https://..."   # custom redirect URL

Custom bangs (standalone shortcuts — no @ support)

[[bangs]]
trigger = "gh"
url = "https://github.com/search?q={}"
name = "GitHub"

# Suppress an auto-generated shortcut
[[bangs]]
trigger = "crates"
enabled = false

Collisions

If two engines produce the same shortcode, SearchHub panics at startup with a message naming both engines. Set shortcode on one of them to resolve it.

Run the web server as a systemd user service

Keeps the web UI running in the background, starts automatically on login.

VERSION=(search_hub --version | cut -d\  -f2)
mkdir -p ~/.config/systemd/user
wget -O ~/.config/systemd/user/search-hub-web.service https://vit.am/~ololduck/search_hub/repository/browse/v$VERSION/contrib/search-hub-web.service
systemctl --user daemon-reload
systemctl --user enable --now search-hub-web.service

Check status with systemctl --user status search-hub-web. View logs with journalctl --user -u search-hub-web -f.

Auto-import with systemd

VERSION=(search_hub --version | cut -d\  -f2)
mkdir -p ~/.config/systemd/user
wget -O ~/.config/systemd/user/search-hub-import.service https://vit.am/~ololduck/search_hub/repository/browse/v$VERSION/contrib/search-hub-import.service
wget -O ~/.config/systemd/user/search-hub-import.timer https://vit.am/~ololduck/search_hub/repository/browse/v$VERSION/contrib/search-hub-import.timer
systemctl --user daemon-reload
systemctl --user enable --now search-hub-import.timer

This imports bookmarks from Zen Browser daily. Edit the file to import from another browser.

Auto-update with systemd

VERSION=(search_hub --version | cut -d\  -f2)
mkdir -p ~/.config/systemd/user
wget -O ~/.config/systemd/user/search-hub-update.service https://vit.am/~ololduck/search_hub/repository/browse/v$VERSION/contrib/search-hub-update.service
wget -O ~/.config/systemd/user/search-hub-update.timer https://vit.am/~ololduck/search_hub/repository/browse/v$VERSION/contrib/search-hub-update.timer
systemctl --user daemon-reload
systemctl --user enable --now search-hub-self-update.timer

This checks for new releases weekly and updates the binary automatically.

Run with Podman / Docker

A container image is available at oci.vit.am/search-hub:latest. It serves on port 8080 as the search_hub user and expects:

• Config mounted at /home/search_hub/.config/search_hub/config.toml
• Database directory mounted at /home/search_hub/.local/share/search_hub/

# Pull and run
podman run -d --name search-hub \
  -p 8080:8080 \
  -v ~/.config/search_hub:/home/search_hub/.config/search_hub:ro \
  -v ~/.local/share/search_hub:/home/search_hub/.local/share/search_hub \
  oci.vit.am/search-hub:latest serve

# SIGHUP reload (re-reads config)
podman kill -s HUP search-hub

# Build locally from the Containerfile
podman build -t search-hub:latest -f Containerfile .

docker-compose

docker compose up -d

See docker-compose.yaml at the project root. A SearXNG service is included as a commented-out example.

Podman Quadlet (systemd-native)

mkdir -p ~/.config/containers/systemd
wget -O ~/.config/containers/systemd/search-hub.container https://vit.am/~ololduck/search_hub/repository/browse/main/contrib/search-hub.container
systemctl --user daemon-reload
systemctl --user enable --now search-hub

The Quadlet file uses %h (your home directory) for volume source paths.

Resources

• Downloads: vit.am/~ololduck/search_hub/latest
• Repository browser: vit.am/~ololduck/search_hub/repository
• Git clone: git clone https://vit.am/~ololduck/search_hub/repository.git search_hub

Changes in 0.7.0

Added

• tagging cargo feature (included by default) gating fastembed/ONNX Runtime. Build with --no-default-features to produce a ~30% smaller binary without semantic auto-tagging.
• Second abbaye cargo builder with no_default_features = true so release page offers both binaries.

Changed

• Self-updater appends -no-default to the download URL when the running binary was compiled without the tagging feature.
• Pruned tokio from features = ["full"] to only needed features (rt-multi-thread, macros, signal, sync, time, fs).
• Pruned actix-web from default features to only macros + compat-routing-macros-force-pub.