Memi
Provides a read-only design engineering audit and skill layer for coding agents.
Memi functions as a read-only design engineering audit and skill layer, equipping AI coding agents like Codex, Claude Code, Cursor, and Grok Build with file-anchored interface evidence before they modify UI. It inspects existing products to pinpoint risks related to accessibility, hierarchy, state, responsiveness, motion, and design tokens. After a fix, Memi can deterministically rerun checks to verify improvements, ensuring the codebase remains the single source of truth for design.
主な機能
01Identifies accessibility, interface craft, product states, and design system risks.
02Offers multiple integration points including CLI, Agent Skills, GitHub Action, and MCP server.
03Enables deterministic design quality checks for CI/CD pipelines.
04Provides file-anchored interface evidence directly to AI coding agents.
05Performs read-only design engineering audits on frontend code.
0629 GitHub stars
ユースケース
01Equipping AI coding agents with design context and verification capabilities for UI development.
02Integrating design quality gates into pull requests for continuous integration.
03Auditing frontend interfaces for design and accessibility risks before making UI changes.