Claude Code Now Supports Artifacts, Live

Anthropic shipped something quietly useful this week.

Claude Code — Anthropic's agentic coding tool — now supports artifacts: live, shareable web pages built directly from your session context. As of June 18, 2026, any Claude Code session can produce a visual page — a PR walkthrough, incident timeline, dashboard, or release checklist — that updates itself as the work continues.

The link stays the same. The page refreshes in place. Everyone looking at it sees the same version.

Claude Code Live Artifacts - What it Does

The value lies less in the page itself and more in what it replaces: the verbal status update. The "let me walk you through what the agent found" Slack message. The screenshot of a terminal that nobody can read.

Claude Code builds the artifact from the full session context — your codebase, connected tools, and the conversation itself. An incident page can surface the failing test, the error spike from a monitoring connector, and the root-cause reasoning from the same session, without you wiring up a single data source. You ask for a page. It builds one from what already exists.

Every publish creates a new version at the same URL, preserving version history.

The Collaboration Problem This Solves

Engineering teams don't lack information. They have a distribution problem. An engineer finishes an investigation. She knows what she found. Her team doesn't — until standup, or until she writes something up, or until she screenshares at an inconvenient hour.

Artifacts collapse that gap. Anthropic's internal testing used incident investigations as the primary use case: kick off a session before standup, publish an artifact, share the link. By the time the meeting starts, the page has already been republished twice with new findings. The team reads the same context. The meeting gets shorter.

That's the actual pitch — not the visual output, but the reduction in communication overhead around agentic work.

Who Artifact Building Is For

Anthropic's announcement lists use cases by role: license audits for legal, data-flow maps for privacy reviews, PR walkthroughs for engineers, UX variations for designers built from real component libraries, and incident postmortems for SREs. The breadth suggests they're positioning this less as a developer tool and more as an organizational one — something that makes Claude Code's output legible to people who aren't running the session themselves.

For marketing and growth teams building on AI infrastructure, this matters more than it might look. As AI agents do more of the actual work, the bottleneck shifts from execution to visibility. Teams need to know what the agent did, why, and what changed. Artifacts are one answer to that.

Availability and Access

Currently in beta for Claude Team and Enterprise orgs, accessible via the Claude Code CLI and desktop app. Pages are viewable in any browser by authenticated org members. Admins control access through role-based scoping, retention policies, and a compliance API. Nothing is public by default.

For teams already running Claude Code, this is worth turning on. For teams evaluating whether agentic coding fits their workflow, this makes the output considerably easier to oversee — which has been one of the stickier objections to putting AI agents into production pipelines.

Want to know how tools like Claude Code fit into a real AI growth strategy? Winsome Marketing helps teams cut through the noise and build systems that work. Let's talk.