What problem does it solve? Agent-generated HTML artifacts like diagrams, plans, and UI mockups are hard to review in plain chat, and human feedback often arrives as vague prose. Galley provides a local review surface where humans annotate artifacts directly, returning source-anchored comments as structured prompts the agent can act on. ## Core Features & Use Cases - Live review sessions: Open or resume a local review surface for an HTML artifact, then block on galley wait until feedback, findings, or session end arrives. - Render audits and playbooks: Run galley check for headless render audits before involving a human, and pull authoring playbooks (diagram, table, comparison, plan, code, input, slides) plus design-system inspection via galley design. - Use Case: You are planning a database migration. Write a plan artifact to .galley/migration.html, run galley check to fix render issues, open the review surface, and wait for the human's anchored comments before implementing. ## Quick Start Ask the agent to create an HTML plan artifact for your current task and open it for visual review with galley, then wait for your inline feedback.