artifact-publish

Publish a self-contained HTML file as a shareable artifact on artifacts.nexaedge.com, and get a link back.

It exists because a self-contained HTML solves "generate" but not "share": sending the file over chat loses the version, and a raw bucket link resolves against the browser's default account, which 403s anyone with more than one.

How access works

An artifact belongs to the domain of the Google account that published it, and only that domain can open it. There is no registration and no allowlist: sign in with your work account and the domain of that account becomes the boundary. Personal accounts (gmail.com and friends) are refused, because treating gmail.com as a domain would mean granting access to everyone with a Gmail.

For people outside the domain, --share N returns a link that opens with no sign-in and expires in N days.

The listing shows only what you published, not the whole domain. There is no index of who uses the service.

Install

/plugin marketplace add nexaedge/nexaedge-marketplace
/plugin install artifact-publish@nexaedge-marketplace

Requires Python 3 (standard library only) and a browser for the first sign-in.

Use

Ask Claude to publish, share or send a link to an HTML file. Or run it directly:

python3 <skill-dir>/scripts/publish.py report.html
python3 <skill-dir>/scripts/publish.py report.html --as acme.com --share 7
python3 <skill-dir>/scripts/publish.py --whoami

Configuration (optional)

~/.config/nexaedge/artifacts.json — every key optional:

{
  "service": "https://artifacts.nexaedge.com",
  "default_domain": "acme.com",
  "accounts": {"acme.com": "you@acme.com"}
}

accounts only preselects the account on Google's sign-in screen. default_domain is the account used without --as; with a single account signed in, that one wins and no configuration is needed.

Sessions live in ~/.config/nexaedge/artifact-share/<domain>.json with mode 600, one file per domain. A refresh token is worth a whole Google account until revoked; revoke at myaccount.google.com/connections.

Quota

Publishing is open to any Workspace domain, so a quota is what bounds it: a new domain starts with a low ceiling on artifact count and total bytes. The refusal says how much is in use. Ask for a larger ceiling at hello@nexaedge.com.

Each file is capped at 100 MB.

Writing the HTML

Prefer self-contained: CSS, JS and data inside the file, images as data URIs. External resources do load — CDN scripts, remote fonts, images by URL, cross-origin fetch all work — so this is a recommendation and not a limit the service imposes. An artifact is a frozen copy meant to open weeks after the link went out, and a CDN that moves takes the page with it, a network that blocks the host renders it wrong, and every open tells whoever serves the asset who is reading.

The isolated origin (sandbox allow-scripts, no allow-same-origin) is what keeps a published document away from this service's own origin and session. Its cost is anything tied to an origin: localStorage, sessionStorage and cookies are unavailable, so a page cannot remember state between visits. The document's own JavaScript runs normally — charts, filters, sorting, canvas, blob downloads.

Give it a real <title>. It becomes the name in the listing and in the browser tab, so index or report makes the artifact unfindable later. Whoever published can fix it afterwards by clicking the title in the top bar.