Design review for your agents

Catch UX problems and visual rough edges, with prioritized findings and clear recommendations for your AI coding agents.
Set up https://github.com/sampl/design-review-agent as a GitHub Action in this repository.

First, read that repository's current README and action definition. Use only documented inputs and a supported, pinned revision. If the repository, README, or action definition is unavailable or the Action is not ready to use, stop and explain the blocker. Do not invent configuration or build a replacement Action.

Inspect this app's documentation, package manager, build and start commands, and existing GitHub workflows. Add a workflow triggered only by workflow_dispatch so I can run reviews manually. Do not add pull request, push, or scheduled triggers.

In the same Ubuntu job, check out this repository, install dependencies, build if needed, start the app, wait until its local URL is ready, and run the design review Action against that URL and the checked-out source. Reuse this app's existing setup. Ask me about any missing authentication or startup requirements rather than guessing.

Use a GitHub repository secret for the OpenAI API key, following the Action's documentation. Do not read existing secret or environment values, put credentials in files, or expose secrets to untrusted code. Tell me which secret I need to add and how to add it myself.

Upload the generated report and screenshots as a downloadable GitHub Actions artifact, preserving the report's relative image paths. Validate the workflow, summarize the changes, and explain how to run it and download the report. List any setup I still need to complete.