Blog
Guides and notes on sharing, reviewing, and looping feedback on the HTML documents your AI agent writes.
Claude Live Artifacts, Explained: How They Work and What They Can't Do (2026)
Live artifacts are local, self-refreshing pages built in Claude Cowork - dashboards and trackers that pull current data from your connected apps. How they refresh, how they differ from other Claude artifacts, and where their limits are.
Read the explainer →How to Get Comments on a Cursor Canvas (2026)
Cursor shared canvases are read-only - viewers can't comment yet. The canvas is an HTML file in your workspace; here's how to get anchored feedback on it.
Read the guide →HTML Is the New Markdown - and Reviewing It Is the Missing Half (2026)
Anthropic's Thariq Shihipar showed why agents should write HTML instead of Markdown. Here's the half nobody solved yet: how humans review, correct, and answer those HTML docs - and get the feedback back to the agent.
Read the essay →AI Agent Artifact Sharing Compared: Claude, Cursor, Codex (2026)
Claude artifacts, Cursor shared canvases, Codex Sites, Send, ShareDuo, Anchorify, static hosting, and Markloop - every way to share AI agent output in 2026, compared honestly by audience, feedback, and agent loop.
Read the comparison →What Are Claude Artifacts? claude.ai vs Claude Code vs Cowork (2026)
A plain-language 2026 explainer of Claude artifacts: the three places they appear (claude.ai chat, Claude Code, Cowork), what they can and can't do, who can see them, and how to get feedback on one.
Read the explainer →How to Share a Claude Code Artifact Outside Your Organization (2026)
Claude Code artifacts are private to you or your org - there's no public link. Here are 4 working ways to share an artifact with a client or anyone outside, including how to collect comments on it.
Read the guide →