"I shipped an open-source CLI to an audience of zero — the most-downloaded build surprised me"

Two months ago I pushed fast-copy to GitHub with no ads, no influencers, no launch budget, and a following I could count on zero hands.

Today the most-downloaded build is the Windows one.

That's the platform that actively fights unsigned binaries — SmartScreen warnings, "run anyway" prompts, Defender getting twitchy about anything that came out of PyInstaller. And it's still on top. I did not expect that.

91 downloads. People found it through Google, DuckDuckGo, Bing — one even came from Yandex. Strangers shared it in Slack channels at companies I've never heard of. No funnel, no campaign. Just search and word of mouth.

The lesson I keep relearning: you don't need an audience to start. You need something that solves a real problem, and then you need to be patient.

What it does today

fast-copy is a file-transfer CLI for people who live in the terminal. The parts that are actually shipped:

Sparse-file-aware backups. This is the one that makes people forward it to a coworker. If you've got a 2.3 TB VM image with 12 GB of real data in it, fast-copy reads the extent map and copies the 12 GB — not the 2.3 TB of zeroes.

$ ls -lh disk.img
-rw-r--r-- 1 root root 2.3T disk.img

# only the real blocks move
$ fast-copy disk.img /backups/disk.img

Server-to-server streaming over SSH. Copy between two remote hosts without pulling everything down to your laptop and pushing it back up. The data streams host-to-host directly.

$ fast-copy src-host:/data/ dst-host:/backup/

Multi-source copies that preserve where each file came from (rsync -R semantics — parent paths are reconstructed at the destination).

$ fast-copy -R /etc/nginx /var/www/html backup-host:/snapshots/
# recreates /snapshots/etc/nginx and /snapshots/var/www/html

Plus block-order copying (offsets sorted for sequential reads), content dedup so identical files don't get copied twice, and post-copy verification so you know it landed intact.

What I'm building next

To be clear: none of this has shipped yet. It's the roadmap, not the changelog.

• Cloud copy to S3, Azure Blob, and GCS — with dedup running server-side, so identical files get copied inside the bucket instead of re-uploaded over your connection.
• An encrypted credential vault (AES-256-GCM) so your cloud and SSH secrets never sit in plaintext. And one I genuinely haven't decided on: a GUI. I built this for people who are comfortable in a shell — but maybe that's leaving people out. If you'd actually use one, that's the signal I need.

The question I'm actually asking

CLI or GUI — which one are you? Drop it in the comments.

If you want to know the moment one of these ships, there's a signup at fast-copy.dev — pick the feature you care about and I'll send a single email the day it lands. Nothing else.

Repo and docs: fast-copy.dev