Can viewers comment on a published Claude artifact?

Not natively — Claude's publish feature lets anyone view and interact with the artifact, but there's no built-in mechanism for viewers to leave pinned notes or comments for the creator. To collect specific, element-level feedback, paste the published artifact URL into a review tool (Drafty, Markup.io) that supports guest commenting on any URL.

Does my client need a Claude account to see the artifact?

No — if you publish it. A published artifact generates a standalone URL that anyone can open in a browser with no account. If you send the conversation URL instead, they'll land inside your Claude chat and will need an account to do anything. Publish first, then send the published link.

How do I share a Claude artifact with a client for review?

Open the artifact panel, click the down caret next to Copy, choose Publish, and copy the resulting URL. Send that link — not the claude.ai/chat URL. Your client opens it in any browser without logging in, sees only the finished artifact, and can interact with it. For pinned feedback, drop that same URL into a review tool before sending.

How do I get specific, element-level feedback on a Claude artifact?

Claude's published artifacts support view and interaction but not native commenting. The workaround: paste the published URL into a review tool that works on any URL. The client opens the review link, clicks the exact element they're talking about — a heading, a button, a chart — and leaves a note pinned there. You get precise feedback instead of "the thing in the middle section."

Can I update the artifact after sharing it and have clients see the new version?

Yes — Claude republishes to the same URL when you continue the conversation and ask for changes. Anyone who already has the link sees the updated version on refresh. One caveat: if you unpublish and republish, you get a new URL. Keep the original published link live while you're in an active feedback round.

What kinds of Claude artifacts can I share for feedback?

Any artifact Claude can publish — HTML apps, React components, markdown documents, SVG images, Mermaid diagrams, and data visualizations. Doc and markdown artifacts work well for report or proposal review. Interactive HTML artifacts work well for prototype feedback. Review tools that accept a URL work on all of them.

drafty

How to get feedback on a Claude artifact

Quick answer

To get feedback on a Claude artifact, publish it first (the share link shows your conversation, not the live artifact), then send clients the published URL — they can view and interact with it without a Claude account. For specific, element-level feedback, drop the URL into a review tool so they can pin a note to the exact button or section they mean instead of describing it in a message.

Step 1

Publish the artifact — don't send the conversation link

Claude's interface has two separate links that look similar but do very different things. The conversation URL (claude.ai/chat/…) lands the client inside your chat history — they see your prompts, your iterations, and every half-built draft along the way. That's rarely what you want a client to see. The right move: open the artifact panel, click the down caret next to Copy, and choose Publish. Claude generates a standalone URL that shows only the finished artifact, with no chat history attached. Clients can open this on their phone or laptop without a Claude account — they just get the doc, the dashboard, or the app, exactly as you built it. This is the step most people skip, and it's why clients come back saying they couldn't find the thing you sent.

Step 2

Drop the published URL into a review tool so clients can pin notes to the exact spot

A published Claude artifact is a live URL — which means you can paste it into any review tool that works on URLs. Tools like Drafty, Markup.io, or Ruttl turn the artifact into a review board: the client opens a new link, hovers the element they're talking about, clicks, and leaves a note pinned right there. No account. No install. The feedback arrives anchored to the exact section, button, or heading they mean. Without this, you're translating "the part near the top of the second section" from a Slack thread into a specific element in your artifact. With element-anchored comments, there's no translation — the note is literally on the thing they meant. Works on doc artifacts, dashboard artifacts, and HTML app artifacts equally well.

Step 3

Walk through it on a screen share for flow and narrative feedback

For anything longer than a single-page doc — a multi-section report, a step-through dashboard, an interactive tool — a 20-minute screen share catches things no annotation tool will. Open your published artifact and share your screen on Zoom or Meet. The one rule that makes screen shares useful: let the client react before you explain anything. Ask "what do you read first on this page?" before you point at the headline. Ask "what would you tap here?" before you walk them through the intended flow. Their first reaction without coaching is the most valuable thing you'll get — it tells you whether the artifact communicates on its own. Take notes live; don't argue or clarify. Async annotation catches specific edits; screen shares catch structural mismatches.

Step 4

Send a structured feedback form for clients who won't open a tool

Some clients won't click a review link no matter how frictionless it is. For those cases, send a short form with a screenshot of the artifact as the header image and three to four concrete questions — one per major section. "Is the recommendation in the Executive Summary clear?" gets a useful answer. "What do you think?" does not. Limit to five questions and include a freeform field at the end. Screenshot-based forms work surprisingly well for doc-style artifacts (reports, proposals, plans) where the client is reading linearly rather than exploring. The tradeoff: you lose element-level specificity and you'll still get "the section near the middle" in freeform answers. Use this as a fallback, not a primary path — async pinned comments produce more actionable feedback per round.

The faster way

If clients keep describing the wrong thing or sending feedback on the wrong version, paste your published Claude artifact URL into Drafty and send the review link. They open it on their phone or laptop, click the exact element they mean, and leave a pinned note there. No Claude account needed — no account of any kind. You see every comment on one artifact, resolve them when fixed, and the updated version lives on the same link. Works on any published Claude artifact URL.

Open a live demo

Questions

Can viewers comment on a published Claude artifact?: Not natively — Claude's publish feature lets anyone view and interact with the artifact, but there's no built-in mechanism for viewers to leave pinned notes or comments for the creator. To collect specific, element-level feedback, paste the published artifact URL into a review tool (Drafty, Markup.io) that supports guest commenting on any URL.
Does my client need a Claude account to see the artifact?: No — if you publish it. A published artifact generates a standalone URL that anyone can open in a browser with no account. If you send the conversation URL instead, they'll land inside your Claude chat and will need an account to do anything. Publish first, then send the published link.
How do I share a Claude artifact with a client for review?: Open the artifact panel, click the down caret next to Copy, choose Publish, and copy the resulting URL. Send that link — not the claude.ai/chat URL. Your client opens it in any browser without logging in, sees only the finished artifact, and can interact with it. For pinned feedback, drop that same URL into a review tool before sending.
How do I get specific, element-level feedback on a Claude artifact?: Claude's published artifacts support view and interaction but not native commenting. The workaround: paste the published URL into a review tool that works on any URL. The client opens the review link, clicks the exact element they're talking about — a heading, a button, a chart — and leaves a note pinned there. You get precise feedback instead of "the thing in the middle section."
Can I update the artifact after sharing it and have clients see the new version?: Yes — Claude republishes to the same URL when you continue the conversation and ask for changes. Anyone who already has the link sees the updated version on refresh. One caveat: if you unpublish and republish, you get a new URL. Keep the original published link live while you're in an active feedback round.
What kinds of Claude artifacts can I share for feedback?: Any artifact Claude can publish — HTML apps, React components, markdown documents, SVG images, Mermaid diagrams, and data visualizations. Doc and markdown artifacts work well for report or proposal review. Interactive HTML artifacts work well for prototype feedback. Review tools that accept a URL work on all of them.

Keep exploring

How to get feedback on a v0 app How to get feedback on a web app How to get feedback on a prototype Design feedback tool

Stop emailing files back and forth.

Share one link. They comment on the exact spot — no account, always the current version.

Open a live demo