Safeguard
Developer Tools•

Claude Artifacts: templates, export formats, storage and sharing

What Claude Artifacts are, which templates exist, what each exports to, the 20 MB storage limit and why artifacts start private, per Anthropic.

Safeguard Security Team
3 min read

Artifacts are standalone outputs Claude creates beside the chat, such as documents, decks, designs, dashboards or small interactive tools. You can edit, revisit and share them. They start private to you, and creating them in chat is available on every plan, from Free to Enterprise.

When Claude makes one

Anthropic's help article says Claude typically creates an artifact for content that is substantial (often over 15 lines), self-contained and likely to be edited or reused outside the chat. Everything saved lives in the Artifacts tab in the sidebar.

Templates and export formats

  • Docs: collaborative documents. Export to Word, PDF, Markdown or Google Docs.
  • Slides: decks you can present. Export to PowerPoint, PDF or Google Slides.
  • Design: visuals and prototypes using your design system. Export as a zip, PDF, PowerPoint or HTML.
  • Dashboards (beta): data driven pages, on Pro, Max, Team and Enterprise.
  • Motion (beta): animated explainers, on Team and Enterprise, exporting as MP4.

Docs, Slides and Design templates are on all plans and on by default. The page says that for Enterprise they turn on by default on October 15, 2026. Dashboards are off by default on Enterprise.

Editing

You can ask Claude for changes or edit template based artifacts directly. In a doc, highlight text and use Edit with Claude. Editing an earlier chat message creates a separate branch with its own artifacts. If an artifact errors, a Try fixing with Claude button exists, though success is not guaranteed.

Connected apps and stored data

Artifacts on Pro, Max, Team and Enterprise can connect apps and store data. They can read from and write to connected apps such as Asana, Google Calendar and Slack, plus custom connectors. Claude asks for approval the first time, and each person connects their own apps.

Data can be saved between sessions in personal storage (visible only to each user) or shared storage (visible to everyone using the artifact). Each artifact has a 20 MB text-only limit. The page advises checking whether an artifact uses shared storage before entering sensitive information.

Artifacts that call Claude

You can build small apps that call Claude. Users sign in with their own Claude accounts, so you do not need API keys, and their usage counts against their own plan limits.

Mobile and Claude Code

On iOS and Android you can view and edit artifacts, but sharing settings change on web or Desktop. Claude Code can publish session output as a live artifact at a private URL.

A practical review step

Before sharing, ask three questions. Who can open it? Does it read from a connected app on behalf of each viewer? Does it write to shared storage? Anything with live connector access deserves the same care as a connector, covered in custom connectors and risks.

Safeguard offers a free scan of code and dependencies at gold.safeguard.sh/scan.

Quick check

Are artifacts public by default?

No. They start private to you.

How much can an artifact store?

Each artifact has a 20 MB text-only limit, according to the help article.

Sources

Last reviewed: 19 October 2026.

Never miss an update

Weekly insights on software supply chain security, delivered to your inbox.

Self-healing security runs on Safeguard.

Your first fix PR is minutes away.

No sales call required, even your agent can complete the purchase over MCP.