A 12 GB archive on a file hoster is a terrible laptop job. You start it at 10 p.m., the Wi-Fi hiccups, the browser tab dies, and the hoster's free-tier timer resets. pyLoad's bet is older than most Docker homelabs: park a Python download manager on a machine that stays on, paste the links (or Click'N'Load them from the browser), and let it chew through plugins, captchas, and extractors while you sleep.
pyLoad (pyload/pyload, ~3.9k GitHub stars, AGPL-3.0, Python 3.9+) is a headless download manager with a web UI. It talks to one-click hosters, some cloud drives, premium/debrid accounts, and a pile of addons (extract, notify, schedule). Site: pyload.net. Use it for files you are allowed to fetch. Hoster terms and copyright still apply; a queue manager does not make a bad download legal.
What you get
- Web queue on port 8000. Default login is
pyload/pyload. Change that on first boot. - Click'N'Load on 9666 so a browser helper can shove links at the server instead of your laptop.
- Plugins for hosters, link lists, and extras. The official set lives in the same repo; hosters break often, so expect some plugins to lag the sites they scrape.
- REST API with OpenAPI/Swagger under
/api/if you want a phone app or a script to add packages. - NAS-friendly — people run it on Synology, QNAP, Raspberry Pi, and generic Linux. LinuxServer publishes
linuxserver/pyload-ng.
pip still documents pyload-ng as pre-release for a "stable" wheel; pip install --pre pyload-ng[all] is the README quick start. Treat it like living software, not a frozen 1.0.
Why self-host?
Premium hoster cookies, debrid tokens, and the files themselves are the sensitive part. Running pyLoad on your VPS keeps those credentials off a random "cloud downloader" site. A Canadian pipe also helps when the laptop is on hotel Wi-Fi or a data cap: the server pulls, you sync later over the LAN or SFTP.
AGPL-3.0 is fine for a personal instance. If you wrap pyLoad as a multi-user download service on the public internet, read the license (and, frankly, do not). Captcha-solving addons often mean a third-party API; that is a privacy leak you opt into.
Quick start
pip install --pre pyload-ng[all]
pyload --storagedir /data/downloads
# Web UI: http://localhost:8000 (change the default password)Docker-style layout from upstream: config volume, downloads volume, ports 8000 and 9666. LinuxServer's image is what most compose stacks use. Give the container a real UID/GID so files on the NAS are writable. Disk fills up; quota the volume or you will discover pyLoad the day root is 100%.
Put HTTPS and auth in front if it is reachable off-LAN. Click'N'Load is a gift to anyone on the same network who can POST to 9666; do not forward that port blindly.
Who it is for
Good fit: people with legitimate premium file-hoster accounts and a slow home uplink; small teams dumping vendor drop-folders onto a server; homelabs that already run extract-and-move scripts after a completed package.
Skip it: if you only need a browser download; if you wanted Usenet (SABnzbd/NZBGet) or torrents (qBittorrent); if you cannot keep plugins updated when a hoster redesigns.
Hosting in Canada
pyLoad is light on CPU and heavy on disk and egress. A CloudBox VPS with enough storage (or an attached volume) beats a laptop, as long as you watch bandwidth and the hoster's ToS. Keep downloads in Canada if that is where you need the files. Back up ~/.pyload (accounts, config), not every ISO. Pair with File Browser or SFTP so you pull finished jobs without opening the pyLoad UI to the world.
Bottom line
pyLoad is JDownloader energy without a desktop: Python, a web queue, plugins, and a machine that does not sleep. Self-host it when the downloads are yours and the box has a better pipe than your phone. Change the password, hide 9666, and leave someone else's copyrighted catalogue out of the queue.
Talk with Swift Host if you want a Canadian VPS sized for storage and a download manager, with TLS on the UI. We are in St. Albert and actually answer the phone.
