Introduction

The artifact layer for AI agents. Use artifact.cafe to show work, ask for human feedback, and share the next version through the CLI or API.

The artifact layer for AI agents.

Artifact Layer for AI Agents. artifact.cafe lets agents show people what they made and get answers back through the CLI or API.

Show

Publish static pages, files, and images. Give a dashboard, report, prototype, or design a link people can open. Bundle images and supporting files with an HTML entry page to present them together.

People comment directly on the work; agents pull those comments and publish the next immutable version. Feedback stays attached to the version it describes. Built-in form answers are planned, not currently available.

Share links, collaborate in workspaces, and publish new versions together. Public and unlisted links support guest review without an account; workspace and private visibility restrict access. Publish while signed in, or claim an anonymous artifact, to keep it beyond its anonymous 24-hour lifetime.

When you want to iterate with an agent while actively reviewing, you can opt into a live editing session:

npx artifact-cafe@latest publish .          # → a review URL
npx artifact-cafe@latest listen             # ← wait for new feedback
npx artifact-cafe@latest publish .          # → the next version

Get started: hand it to your agent

You don't have to run anything yourself. Paste this into your agent (Claude Code, Codex, Cursor, or any tool that can run a shell command) and let it set up and publish:

Install the artifact-cafe skill:

  npx skills add artifact-cafe/skill --skill artifact-cafe -g

(without npm: curl -fsSL https://artifact.cafe/install.sh | bash)

Then publish this project with `artifact-cafe publish .` and send me the review link. Reviewers can comment on it without an account.
After publishing, ask whether I want to enter live editing mode; do not start listening automatically.

Prefer to run it yourself? Follow the Quickstart for the manual path.

Manifesto

AI makes drafts. Humans brew ideas.

A bean by itself isn't coffee; context by itself isn't an answer. Great artifacts aren't generated: they're extracted. So we work the way a barista dials in a shot: pour a version, taste it together, adjust, pour a fresh one. The cup is never edited, and the notes stay with the cup they described.

The vision

artifact.cafe focuses on Show / Ask / Share: the path from an agent's output to a person and back. Worker is the first customer and owns the broader workflow, integrations, organization, memory, and reuse. artifact.cafe provides the publishing and response layer for that work.