share

CLI file sharing and link shortening with download tracking. Backed by Cloudflare R2 + Workers KV + Workers. Zero cost, full ownership.

Live at icecube.to · 🧊.to

Why

Google Drive has no CLI. transfer.sh is dead. Presigned S3 URLs expire. This uploads a file or shortens a URL, gives you a short link, copies it to clipboard. Done.

$ share upload README.md --slug readme.md --public
https://icecube.to/readme.md (copied)

$ share link https://github.com/adriangalilea/share --slug gh --public
https://icecube.to/gh → https://github.com/adriangalilea/share (copied)

$ share ls
                              Files
┏━━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━┳━━━━━┓
┃ Slug      ┃ Name      ┃     Size ┃ Uploaded   ┃ DLs ┃ Vis ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━╇━━━━━┩
│ readme.md │ README.md │   7.2 KB │ 2026-03-19 │  15 │ pub │
└───────────┴───────────┴──────────┴────────────┴─────┴─────┘
                              Links
┏━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━┳━━━━━┓
┃ Slug ┃ URL                                       ┃ Created    ┃ Clicks ┃ Vis ┃
┡━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━╇━━━━━┩
│ gh   │ https://github.com/adriangalilea/share     │ 2026-03-19 │      2 │ pub │
└──────┴───────────────────────────────────────────┴────────────┴────────┴─────┘
icecube.to/<slug>

$ share rm gh
Deleted https://github.com/adriangalilea/share (/gh)

This README is shared at icecube.to/readme.md.

Usage

share upload <file>                         # upload file, auto-generate slug
share upload <file> --slug <name>           # custom slug
share upload <file> --public                # show on landing page (default: private)
share upload <file> --name <name>           # override download filename
share upload <file> --keep-metadata         # skip EXIF/metadata stripping
share link <url>                            # shorten a URL
share link <url> --slug <name>              # custom slug for short link
share link <url> --public                   # show on landing page
share ls                                    # list all files and links
share rm <slug>                             # delete by slug, filename, URL, or r2_key
share setup                                 # interactive first-time config

Uploads strip EXIF/metadata by default (images via Pillow, videos via ffmpeg).

Install

uv tool install git+https://github.com/adriangalilea/share.git

Architecture

CLI (Python)                        Cloudflare Free Tier
┌──────────────┐    boto3/S3 API    ┌─────────────────┐
│  share CLI   │ ─────────────────→ │       R2        │ files (10GB free)
│              │                    └────────┬────────┘
│              │    CF Python SDK           │
│              │ ─────────────────→ ┌────────┴────────┐
│              │                    │       KV        │ slug → metadata
└──────────────┘                    └────────┬────────┘
                                            │
Browser                                     │
┌──────────────┐                    ┌────────┴────────┐
│  short link  │ ─────────────────→ │     Worker      │ serves files,
│              │                    │ (custom domain) │ tracks downloads,
└──────────────┘                    └─────────────────┘ landing page

The CLI uploads files to R2 and writes metadata to KV keyed by slug:<slug>. The Worker on your custom domain looks up slugs, streams files from R2 with range request support (for video/audio streaming), and increments download counters.

Every response carries a Content-Disposition header with the original filename so downloads keep their extension regardless of browser or OS (the slug in the URL has no extension; without this header some browsers save the file extensionless). Previewable types — text/*, anything containing json/xml/javascript, image/*, audio/*, video/*, and application/pdf — are served inline so the link still renders in-tab. Everything else (all archives: .7z, .zip, .tar, .gz, .rar, …; binaries; unknown application/octet-stream) is served attachment and downloads with the correct name.

Self-hosting

Prerequisites

• Python >=3.12, uv
• Cloudflare account (free tier)
• Node.js (for worker deployment)
• A domain pointed to Cloudflare nameservers

1. Install the CLI

uv tool install git+https://github.com/adriangalilea/share.git

2. Create Cloudflare resources

You need two API tokens:

R2 API token (S3-compatible, for file uploads):

• Cloudflare dashboard → R2 → Manage R2 API Tokens → Create
• Permissions: Object Read & Write
• Save the Access Key ID + Secret

CF API token (for KV + worker deployment):

• My Profile → API Tokens → Create Token
• Use template: Edit Cloudflare Workers
• Add permissions: Zone DNS Edit + Zone Read
• Save the token

3. Run setup

share setup

This verifies credentials, creates the R2 bucket and KV namespace, and writes config to ~/.config/share/config.toml.

4. Configure and deploy the worker

Clone the repo and edit worker/wrangler.toml:

name = "share"
main = "src/index.ts"
compatibility_date = "2026-03-19"

[[r2_buckets]]
binding = "R2"
bucket_name = "share"

[[kv_namespaces]]
binding = "KV"
id = "your-kv-namespace-id"  # printed by share setup

[[routes]]
pattern = "yourdomain.com"
custom_domain = true

[vars]
SITE_NAME = "yourdomain.com"

Then deploy:

cd worker
npx wrangler deploy

5. Add your domain to Cloudflare

If not already done:

1. Cloudflare dashboard → Add a site → your domain
2. Update nameservers at your registrar to the ones Cloudflare gives you
3. The worker deploy will create the DNS records automatically
4. SSL/TLS → Edge Certificates → Always Use HTTPS → ON

6. Upload

share upload myfile.pdf
# https://yourdomain.com/kX9mT (copied)

Config

~/.config/share/config.toml — see config.example.toml.

Cloudflare free tier limits

VPN note

R2's S3 endpoint uses TLS that some VPNs break. If uploads fail with SSL errors, bypass VPN for r2.cloudflarestorage.com.

KV schema

slug:<slug> → file: { type?, name, size, content_type, uploaded_at, downloads, r2_key, slug, public }
slug:<slug> → link: { type: "link", url, slug, created_at, clicks, public }