Anthropic has added a feature called Artifacts to its Claude Code offering for Team and Enterprise subscribers. The update converts a Claude Code session into a live, interactive HTML page that can be shared via a single URL, letting engineers and non‑technical colleagues view dashboards, app mockups, or other visual outputs as the AI runs code or pulls data.
Artifacts are built directly from the session’s context, using the local repository, monitoring tools, and conversational reasoning to generate a self‑contained page. The page refreshes in place as the AI updates charts or text, and each change creates a version history that can be rolled back. To keep corporate data secure, the HTML is capped at 16 MiB, all scripts and styles are inlined, and a strict Content Security Policy blocks external network requests.
The move follows OpenAI’s recent rollout of Codex Sites, which provides full‑stack, persistent web apps with database and storage integrations. Anthropic’s approach deliberately avoids backend services, positioning Artifacts as a secure, ephemeral alternative for status reports, whiteboard diagrams, and data visualizations rather than a production SaaS platform.
Lead engineer Boris Cherny highlighted the utility, noting that Artifacts let him share live code explanations, system diagrams, and dashboards instantly with teammates. Enterprise admins retain control through role‑based access and retention policies, suggesting the feature could streamline developer‑to‑stakeholder communication across many organizations.
