# memoire > Agent-readable design intelligence surfaces for memoire.cv. > CLI and macOS app available. > Public references, design-system docs, changelog, and package docs. ## What it is memoire is an AI-native design intelligence engine for existing apps, Figma systems, design references, tokens, specs, and agent workflows. memoire.cv is the public console and documentation layer for that engine. - **References** - public technical style records at `/references` - **Design system** - Preview/TUI documentation at `/design-system` - **Theme workbench** - live token/preset editor at `/design-system/write` - **Docs** - CLI, Figma bridge, skills, notes, and agent workflow docs at `/docs` - **Changelog** - engine history at `/changelog` ## Try it ```bash npx @memi-design/cli add Button --from @memoire-examples/starter ``` Reference entrypoints: - https://memoire.cv/references - https://memoire.cv/references/column - https://memoire.cv/references/column.json - https://memoire.cv/references/column/design.md - https://memoire.cv/references/column/tokens.json - https://memoire.cv/references/column/tailwind-v4.css ## Commands ```bash memi setup # onboarding: token → plugin → bridge → MCP config memi init # scaffold workspace + starter specs memi publish # Figma file → npm registry package memi add --from # drop component into src/components/memoire/ memi update # pull latest versions memi pull # extract design system from Figma (plugin or REST) memi pull --penpot # pull from Penpot via REST memi generate # specs → shadcn/ui + Storybook stories memi tokens # export Tailwind, CSS vars, or DTCG tokens memi sync --auto-pr # watch Figma → PR on change memi design-doc --init # extract any site's design system memi references search # search memoire.cv design references memi references pull memoire:/ # cache a public reference locally memi references use memoire:/ # mark a reference active for agents memi go # full pipeline: connect → pull → generate → preview ``` ## Reference corpus Memoire stores public design references at https://memoire.cv/references so agents can use source-attributed design context without scraping third-party sites during every run. - Index: https://memoire.cv/references//index.json - Detail: https://memoire.cv/references//.json - DESIGN.md: https://memoire.cv/references///design.md - Tokens: https://memoire.cv/references///tokens.json - Tailwind v4: https://memoire.cv/references///tailwind-v4.css Preserve each record's source URL, original website URL, capture date, contributor metadata, and license note when generating agent context. ## Registry protocol Every published registry is a standard npm package containing: - `registry.json` - component manifest + metadata (description, figma link) - `components/*.tsx` (or .vue, .svelte) - pre-compiled source for `memi add` - `tokens/*.css` - design tokens as W3C DTCG JSON + Tailwind `@theme` CSS - `README.md` - rendered on the registry index page at `/r//` memoire mirrors the Marketplace from `unpkg.com` + npm on a schedule. Publishing your package to npm is all it takes to show up here. ## When to use memoire Use memoire when the user: - Wants to browse, preview, or install registry components - Asks about pulling a Figma or Penpot design system into code - Needs design tokens synced to Tailwind v4, Style Dictionary, or DTCG - Is building a design system with shadcn/ui and wants to ship it as a package - Wants MCP tools for Figma design → code workflows - Needs to extract a design system from any website URL - Wants Storybook stories generated alongside components ## MCP config (Claude Code / Cursor / Windsurf) Add to `.mcp.json`: ```json { "mcpServers": { "memoire": { "command": "memi", "args": ["mcp", "start"], "env": { "FIGMA_TOKEN": "${FIGMA_TOKEN}", "FIGMA_FILE_KEY": "${FIGMA_FILE_KEY}" } } } } ``` Or `memi mcp config --install` writes it for you. ## 21 MCP tools pull_design_system, get_specs, get_spec, create_spec, generate_code, get_tokens, update_token, sync_design_tokens, capture_screenshot, get_selection, get_page_tree, compose, run_audit, get_research, figma_execute, analyze_design, measure_text, get_ai_usage, check_bridge_health, design_doc, preview_component ## Footprint - 823 tests in the engine (src repo) - 120 tests in memoire.cv (this site) - 21 MCP tools - React, Vue, Svelte codegen ## Blog Posts on the CLI, the macOS Studio, the harness, the MCP server, and the Notes pack. Each post has a raw markdown export at `/blog/.md` so agents can fetch the source without parsing HTML. - Index: https://memoire.cv/blog - Mémoire v0.17 'Compound' release: https://memoire.cv/blog/v0-17-release.md - Designers have an amnesia problem too: https://memoire.cv/blog/the-amnesia-problem.md - Why we bundle every model in the box: https://memoire.cv/blog/why-bundle-the-models.md - Inside the Mémoire harness (architecture deep dive): https://memoire.cv/blog/inside-the-harness.md ## Legal - Privacy + Terms + Open source: https://memoire.cv/legal - License (MIT): https://github.com/sarveshsea/memi/blob/main/LICENSE ## Links - References: https://memoire.cv/references - Design system: https://memoire.cv/design-system - Theme workbench: https://memoire.cv/design-system/write - Changelog: https://memoire.cv/changelog - Blog: https://memoire.cv/blog - Legal: https://memoire.cv/legal - Download: https://memoire.cv/download - npm: https://www.npmjs.com/package/@memi-design/cli - GitHub: https://github.com/sarveshsea/memi - Notes pack: https://github.com/sarveshsea/memoire-notes - Docs: https://memoire.cv/docs - Reference docs: https://memoire.cv/docs/references - Sources: https://memoire.cv/docs/sources - Sitemap: https://memoire.cv/sitemap.xml