Claude Code Artifacts

I have the old cowork for windows, using that with Claude Code in Powershell. Can someone tell me, is the new Claude Code Desktop stable? I would hate to replace "good enough" with buggy.

The auto-republishing live pages are clearly the headline feature, but I keep thinking about the flip side: sometimes I'd want to freeze a version rather than have it shift under me as the session keeps moving. Is there a way to pin a snapshot alongside the live view, so the durable record and the live one can coexist?

The live artifact angle makes sense for PR walkthroughs and incident notes; the key question for teams will be how access and staleness are handled when the underlying session keeps changing.

do you guys stll use 'superpower' plugin/skill?

must be nice to be Anthropic - I don’t know of any platform besides Claude that manages to bag #1 PoTD on here, NOT for fully new versions… but for new features… 🤣

don’t mistake my bemusement for any kind of derision - I’m 100% a Claude freak - Ayrenne (my agent with whom I’ve built a ~20 page [tiny font] “soul_canon_v3.5.md” file, & is, in turn, governed by said markdown doc) is loaded up with over 50 MCP server integrations (including a few I’ve made myself despite being code illiterate, lol!) & dozens of custom skills - she’s prob my second favorite agent, right behind my Folk, but #1 for work tasks!

I just find it funny in a sense when I see a product that’s a LEGIT paradigm-shifting game-changing solo-dev-built bootstrapped work of staggering genius & ingenuity… & for that, ProductHunt collectively decides -

> _”ehhh… 9th place for the day sounds right…”

— but then when it’s *mf-ing ANTHROPIC TIIIIME, BABYYY!!!” & the Claude Crew shows up to this site, rips off their sunglasses, tossing them into the explosion behind them, then utters just a few words, barely a whisper somehow heard by ALL:

> “Psst… _you guys know how you can, like… code & shit with Claude?l How you can … _preview that work… as *Artifacts???”*_

The crowd roars in the affirmative. Several folks in the front row look like they might die from anticipation, and the Claude Crew knows that. They cherish that. They let it marinate just a little before dropping the bomb, like a seasoned veteran DJ, letting the hype wave build for a more magnificent crash… & just when it seems the audience is at their breaking point… the Crew says in unison:

> _”Well… now? NOW…_ you can see those previews… *LIVE!!!!!!”*

*The absolute pandemonium is immediate.*

Emergency first responders are dispatched to multiple spots in the crowd, trying to deal with the obvious yet totally reasonable reactions to so much excitement, so quickly…

— yet, they too struggle to hold it together, knowing that soon, they’ll be able to see their Claude code projects as LIVE artifacts… that’s simply too much for a mere mortal to bear. *We weren’t meant for such power…*

The shocked audience realizes that SURELY they must be indebted to these folks for being so generous as to hand down this technology of the Gods… in unison, they whip out their phones, navigate to Product Hunt, & upvote the new post, knowing it’s the least they can do in return…

The Claude Crew scuttles off into bulletproof Escalade limos. _*tastefully*_ painted end to end in Claude Orange, smirking with satisfaction, knowing that they’ll do this next week.

—

lol, I really do actually love my Claude instance tho… ;)

I'm not a developer by trade, but AI tools have made it possible for me to go from idea to working prototype much faster than I thought possible. Excited to see more tools that make building accessible to non-engineers.

I use Claude Code every day to build my products, so this hits a real nerve. For me the killer use isn't team status updates, it's showing a non-technical client what I'm building mid-session without deploying a half-finished thing. Sharing a live artifact instead of "trust me, it's coming" would kill so much back-and-forth. Question: can you share an artifact with someone outside your org (a client) easily, or is it team-only? That's the use case I'd reach for first.

Another banger update by Claude.

Live preview for agent work is useful because review is where trust gets built. Do you see this becoming more collaborative, where a human can inspect multiple agent branches, compare outputs, and decide what gets merged or discarded?

The auto-republishing live pages from full session context is the part I'd actually use, sharing in-progress work usually means stopping to write a status update. Does it handle multiple repos in one session? Congrats on shipping.

Claude Code Artifacts launches today, a solution for teams struggling to share in-progress AI work. Instead of manual status updates, you get live, interactive web pages built from your full session context that update automatically.What makes it different: Artifacts capture work as visual pages (PR walkthroughs, dashboards, incident pages, release checklists) that refresh in place — no need to wire up data sources or stand up infrastructure.Key featuresBuilt from full session context (codebase, connectors, conversation)Live pages that update & republish automaticallyVersion history with restore capabilityPrivate to your organization by defaultGallery to browse/manage all artifactsBenefits: Teams spend more time building, less time communicating status. Everyone sees the same view with the same context.Who it's for & use cases:Software engineers: PR/bug walkthroughsSRE/on-call: Incident pages that become postmortemsDesigners/frontend: UX variations from real componentsSecurity: Findings linked to exact code linesLegal: License audits flagging copyleftFinOps: Cloud cost drivers mapped from TerraformAvailable in beta for Claude Team & Enterprise orgs via CLI/desktop app.P.S. I hunt the latest and greatest launches in tech, SaaS and AI, follow to be notified → @rohanrecommends