Anthropic's Claude Code Turns Any Coding Session Into a Live Shareable Web Page
Claude Code can now publish live, auto-updating web pages from your coding session and share them privately with your team
Takeaways
- New feature: Claude Code can now generate live, shareable web pages (Artifacts) directly from your coding session, no extra tooling needed.
- Auto-updating: Artifacts refresh automatically as your session progresses; everyone with the link always sees the latest version.
- Org-scoped privacy: Artifacts are private by default and can only be shared within your organization, not made public.
- Full session context: Pages draw on your codebase, MCP connectors, plugins, and full conversation history, not just a snapshot.
- Version history included: Every publish creates a new version at the same URL, with a gallery to browse and restore past versions.
- Availability: Currently in beta for Team and Enterprise plans via CLI and desktop app.
Claude Code just gained the ability to turn any active session into a shareable, interactive web page. Called Artifacts in Claude Code, the feature lets you generate live pages, think PR walkthroughs, incident dashboards, release checklists, or architecture maps, directly from your coding session, and share them with teammates via a private link. No separate tooling, no manual copy-paste, no standing up infrastructure.
The key word here is live. As your Claude Code session keeps working, the artifact refreshes automatically. Everyone with the link is always looking at the latest version, not a stale screenshot you exported 20 minutes ago.
What makes these different from regular artifacts
Claude has had Artifacts for a while now, as a way to render code, React components, SVGs, and documents in a side panel during chat. But Artifacts in Claude Code are a distinct, more powerful thing. They are built on the full context of your Claude Code session: your codebase, connected plugins, MCP tools, and the entire conversation history.
That means a single artifact can pull together:
- The failing test and the function behind it, straight from your repo
- An error spike from a connected monitoring tool (via MCP connector)
- The root-cause reasoning Claude worked through in the session
You do not wire up data sources manually. You ask for a page, and Claude builds it from what already exists in the session.
The sharing model is org-scoped by design
Every artifact is private to its author by default. When you are ready, you can share it with teammates directly from the page header, and artifacts are viewable only by authenticated members of your org and cannot be made public.
Admins manage access with an org-level toggle and role-based scoping, set retention policies, and get org-wide visibility through the compliance API. This is a meaningful distinction from the regular Artifacts feature, where sharing is free and recipients do not need a Claude account to view published artifacts. In Claude Code, sharing is intentionally locked inside your organization.
Version history, not just a snapshot
Every publish is a new version at the same link, with version history so you can restore at any time, and a gallery lets you browse and manage all artifacts you have made. This is closer to a living document than a one-time export. The URL stays stable while the content evolves underneath it.
The internal debugging example Anthropic shared illustrates the workflow well: an engineer starts an incident investigation before standup, Claude publishes an artifact with a timeline, suspect commits, and an error-rate chart, the engineer shares the link with the team, and by the time standup begins, Claude has republished it twice as the investigation progressed. With artifacts, team members and stakeholders do not have to "walk us through what the agent found" because they are all looking at the same view, with the same context.
Who this is actually for
Anthropic has been explicit about the range of roles this targets. Some of the most compelling use cases from the official blog post:
- SREs and on-call engineers: An incident page that grows as you investigate and becomes the postmortem automatically
- Engineering managers: A page of what actually shipped, built from merged PRs, grouped by project
- Security teams: Audit findings linked to exact lines of code, so the fix is unambiguous
- Staff engineers: A service map drawn from the actual import graph, not a whiteboard sketch
- Legal and open source: A license audit of every dependency, flagging copyleft, straight from the repo
- Frontend engineers: Multiple UX variations of a screen, built from real components so whichever you pick is shippable
The real shift here
The broader pattern Anthropic is pushing is that AI coding agents should not just produce code, they should also produce communication artifacts around that code. Status updates, walkthroughs, audit trails, and dashboards have always been the tax on engineering work. They take time, go stale fast, and require someone to maintain them. Artifacts in Claude Code makes that output a byproduct of the session itself.
This also signals something about where Claude Code is headed as a product: away from a purely terminal-centric tool and toward something that bridges the gap between the engineer running the agent and the broader team watching the work happen. A Claude Code session can range from investigating an incident to refactoring a service to analyzing months of data, and artifacts translate that work into a web page anyone can open and explore.
Availability and access
Artifacts in Claude Code is available in beta to Claude Team and Enterprise orgs, from the Claude Code CLI and desktop app, with pages viewable in any browser. Claude Code is included with every Team plan seat, with premium seats offering more usage for heavier workloads. For Enterprise plans, Claude Code is included with the single Enterprise seat on new and self-serve plans.
You can get started by simply asking your session for an artifact. No special command syntax required. Just describe what you want, and Claude Code builds the page and gives you a link you can open in your browser or the Claude Code desktop app.