← Blog

How to Share a Claude Code Artifact Outside Your Organization (2026)

Short answer: you can't do it on claude.ai. Claude Code artifacts are private to you on Pro and Max plans, and on Team and Enterprise plans they can be shared only with people inside your claude.ai organization. There is no public link and no way to add an outside viewer. To get an artifact in front of a client or anyone external, you have to take the underlying HTML file out - Anthropic's docs say exactly that - and share the file through another channel. The four workable ways to do that are below.

What the docs actually say

An artifact in Claude Code is "a live, interactive web page that Claude Code publishes from your session to a private URL on claude.ai". The official documentation is unambiguous about the sharing boundary:

"Sharing stops at your organization. Viewers must sign in to claude.ai as a member of the same organization that published the artifact, and there is no option to make an artifact viewable outside it. To send the underlying content to someone outside your organization, ask Claude for the HTML file and share that file directly."

Two more limits matter if your reviewer is a client or an external expert:

  • Viewers can't comment. "Artifacts are viewable, not co-edited." There is no feedback mechanism at all - if a teammate spots a problem, that conversation happens somewhere else.
  • Links can expire. Org admins can set retention policies that auto-delete artifacts, so an artifact URL is not a durable deliverable.
PlanWho can see your Claude Code artifactPublic link?Comments?
Pro / MaxOnly youNoNo
TeamPeople in your org you share withNoNo
EnterprisePeople in your org (if an Owner enabled artifacts)NoNo
Claude Code artifacts stop at the org boundary The HTML file doesn't have to. your claude.ai organization claude.ai/code/artifact/… Deploy-failures dashboard private to you (Pro/Max) · shareable only in-org (Team/Enterprise) viewers can look - nobody can comment C client E domain expert T contractor artifact URL the HTML file → Markloop → comments back docs: “ask Claude for the HTML file and share that file directly”
Claude Code artifacts stop at the org boundary - the HTML file doesn’t have to.

First, get the HTML file out

Every option below starts the same way. An artifact starts as an HTML or Markdown file in your project directory - Claude writes it there before publishing, and the page shell is added at publish time. If the source is Markdown - or you're not sure - just ask in your session:

Save the artifact as a standalone HTML file in the project root.

Because the published page is self-contained by design (the artifact CSP forces inline CSS/JS and embedded images), the exported HTML renders identically anywhere. Now pick a channel:

Option 1 - Email the file (the official workaround)

Attach the HTML file and send it. This is what the docs suggest, and it works for a one-shot "take a look".

Good: zero new tools. Bad: some mail clients block or mangle HTML attachments; you've handed over your source file; there are no versions, and feedback comes back as prose in a reply thread that you'll re-type into your agent by hand.

Option 2 - Static hosting (GitHub Pages, Netlify, Vercel, S3)

Drop the file on any static host and share the URL.

Good: free, durable, you control the domain. Bad: the page is public to anyone who finds the URL, updating means redeploying, and there's still no way for the reviewer to comment - you've solved viewing, not feedback.

Option 3 - Artifact-hosting tools (Send, ShareDuo, Anchorify)

A small ecosystem of tools now exists specifically to host Claude artifacts on an external link - Send, ShareDuo, Anchorify. Some add viewer analytics (who opened it, when) or custom domains, which is genuinely useful for sales-style tracking.

Good: purpose-built, quick, works regardless of your Claude plan. Bad: they solve distribution, not review. The viewer still can't comment on a specific paragraph, and nothing flows back into Claude Code. If your goal is sign-off or feedback - a spec a client must approve, a report an expert must correct - you're back to email prose. We compare all of them in detail in Claude artifact sharing options compared.

Option 4 - Markloop (when you need feedback, not just a link)

Markloop is a web app for reviewing agent-made HTML documents - which is exactly what an artifact is. The flow:

  1. Get the file in. Upload the HTML, or skip the file dance entirely: Claude Code pushes it over MCP - one prompt, no manual upload. Codex works too.
  2. Share it. Send one link - no login needed, with optional expiry and version history. Links are view-only by default; flip on "allow comments" and anyone with the link can comment too. Or invite your client by email as a Viewer - either way, no Claude account, always free.
  3. Collect anchored feedback. Reviewers comment on the exact element or sentence, and answer the questions you embedded in the doc. They see the rendered page, not your source file.
  4. Pull it back into your agent. Claude Code pulls every open thread over MCP as structured context - target, quote, intent, version - applies the changes locally, and publishes v2. Comments get marked "addressed"; open questions carry over to the next round.

The honest boundaries: Markloop is for self-contained HTML documents, not live multi-page apps; link commenting is an opt-in toggle (links stay view-only until you enable it); and it's a paid tool for the creator - free 14-day trial, no card required, then $19/month - while reviewers are always free.

The Markloop loop Add → Share → Collect → Pull → Apply ↻ your coding agent Claude Code · Codex 1 · writes the doc (HTML) 2 · markloop_put_version() 4 · markloop_pull() 5 · applies → publishes v2 ✓ order-sync.html v1 → v2 3 · People review 1 “What if the reservation times out?” answered ✓ 1 open → next round push · MCP pull · MCP feedback comes back structured - target, quote, intent, version - and it repeats Reviewers need no account and are always free · the agent applies changes locally - nothing runs on Markloop's servers
The artifact leaves the org; the feedback comes back to your agent - structured and ready to apply.

Which one should you pick?

Email the fileStatic hostingSend / ShareDuo / AnchorifyMarkloop
Works outside your org
Viewer needs no account
Anchored comments on the doc✓ (invite, or link with comments on)
Feedback flows back into Claude Codemanualmanualmanual✓ MCP pull
Versions with "addressed" tracking
Viewer analytics✓ (Send)
Reviewer sees source fileyesyesvariesno - rendered view only

Rule of thumb: one-shot look → email or a static host. Sales-style tracking → an artifact-hosting tool. Sign-off, corrections, or answers you'll feed back to the agent → Markloop.

FAQ

Can I make a Claude Code artifact public?

No. As of July 2026 there is no public-link option for Claude Code artifacts on any plan. (Artifacts made in claude.ai chat are a different feature - Free/Pro/Max users can publish those publicly, but Team/Enterprise chat artifacts also stay inside the org.)

Will the artifact still work as a standalone HTML file?

Yes. The artifact CSP forces everything to be self-contained - inline CSS and JavaScript, images as data URIs - so the exported file renders identically in any browser, including inside Markloop's viewer.

Does my client need a Claude account to review it in Markloop?

No. Invite them by email, or just send the share link with comments turned on - they comment in the browser, no login. No Claude account, no Markloop payment - reviewers are always free.

What happens when I publish a new version?

In Markloop each upload (or MCP push) becomes the next version of the same document. Comments stay attached to the version they were made on, resolved threads are marked addressed, and open questions carry into the new round.

Made an artifact someone outside your org needs to see - and mark up?

Push it to Markloop from Claude Code with one MCP call, send the link with comments on, and pull their feedback straight back into your session.

Start free - 14-day trial, no card required

Reviewers are always free.