# Rubric Rubricは、MCP経由でコーディングエージェントに提供される、厳選されたUXコーパスです。文書化されたUXルールに沿った設計を支援し、同じルールで公開ページを監査して、問題点や深刻度、修正案を提示します。 - Findings24: - ウェブサイト: - 料金: Freemium - カテゴリー: AIデザインアシスタント ## Rubricとは？ Rubric serves a curated UX corpus to coding agents over MCP. It is designed to help agents work from specific UX rules rather than general design advice, so they can make decisions based on documented guidance. The same corpus is used before and after launch: agents can pull guidelines while designing, then audit a live site against the same rules and produce findings with severity, evidence, and suggested fixes. ## Rubricでできること ### Curated UX corpus over MCP Provides a specific corpus of UX rules that agents can query through MCP instead of relying on broad, generalized design advice. ### Design-time guideline retrieval Lets agents fetch prescriptive UX guidance before code is built, including what the rule is, why it matters, and what breaks without it. ### Live-site auditing Evaluates a real page against the same checklist used at design time and returns findings ranked by severity. ### Structured findings output Returns findings as separate blocks that include the issue, severity, impact, actions, and references rather than a single paragraph summary. ### Streamable HTTP MCP server Runs as a standard MCP server over streamable HTTP, which the site says works with any MCP client that speaks the protocol. ### Claude Code plugin support Includes a plugin flow that installs the MCP connection and supports UX audit and reporting tasks from Claude Code. ## 利用シーン ### Guide a new build with UX rules Use Rubric before implementation to pull relevant guidance for a component or flow, so the agent designs to the corpus rather than guessing from general UX advice. ### Audit a shipped checkout or form flow Run a live page through the checklist after release to identify issues, rank them by severity, and generate a report that can be handed to the team fixing them. ### Support automated QA in CI or scheduled jobs Use the HTTP server with a bearer token when no browser can open, such as in CI or a cron job, to run the same audit workflow in automated environments. ### Standardize review across MCP tools Connect Rubric from an MCP-compatible client such as Claude Code, Codex, Cursor, Windsurf, Zed, Cline, Goose, or Manus to keep the same UX rules available across tools. ## よくある質問 ### What does Rubric do? Rubric provides a curated UX corpus to coding agents over MCP, then uses the same corpus to audit live pages against the same rules and report findings with severity and fixes. ### How do you connect it to an agent? The site says you can add it as an MCP server over streamable HTTP, or install the Claude Code plugin that wires up the connection for you. ### Does it work only with Claude Code? No. The site says any MCP client works, and it lists Claude Code, Codex, Cursor, Windsurf, Zed, Cline, Goose, and Manus as compatible clients. ### What does a finding include? A finding is presented as separate blocks, including the issue, severity, impact, actions, and references. ### Is there a free plan? Yes. The pricing page shows a free tier, a pay-as-you-go usage-based plan, and an enterprise option with invoicing and volume terms. ## クイック情報 - **Category**: Developer Tool - **Primary use**: UX guidance and auditing for coding agents - **Protocol**: MCP over streamable HTTP - **Setup options**: Direct server connection or Claude Code plugin - **Pricing model**: Free tier, usage-based pay-as-you-go, and enterprise - **Source domain**: rubric.design