Agent skill

Loora Design Guide

by lassejlv in lassejlv/loora

Build, edit, refine, troubleshoot, and review polished responsive product interfaces through the Loora MCP server and its structured Canvas schemas.

AGPL-3.0Auto-check passedFrontend & Design

Install Loora Design Guide

skills CLI
$ npx skills add lassejlv/loora --skill loora-design-guide -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install lassejlv/loora loora-design-guide --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/lassejlv/loora.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/loora-design-guide .claude/skills/loora-design-guide && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
loora-design-guide
GitHub stars
140
Token cost
~2.4k tokens
SKILL.md length
1,192 words
Files
7 (incl. references)
Skills in repo
1
Repo updated
First seen
Licence
AGPL-3.0

At a glance

Build, edit, refine, troubleshoot, and review polished responsive product interfaces through the Loora MCP server and its structured Canvas schemas.

  • Works in 9 steps: Check capability and orient. Confirm the… → Protect the target. Confirm whether the… → Form a visual direction. Extract the… → …
  • An agent must create
  • SKILL.md covers Gate the MCP surface first, Load references by action, Follow the core loop and Preserve Canvas semantics, plus 2 more sections
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Loora Design Guide is an agent skill from lassejlv/loora. Build, edit, refine, troubleshoot, and review polished responsive product interfaces through the Loora MCP server and its structured Canvas schemas. Use when an agent must create or modify a Loora design, recover from MCP schema errors or missing authoring tools, turn a brief or reference into editable Canvas nodes, establish tokens or reusable components, add themes, interactions, or motion, work safely on a Loora branch, verify with or without image vision, or export implementation code without treating HTML…

Its SKILL.md is about 2.4k tokens, which your agent loads only when the skill is triggered. The skill folder holds 8 other files, including reference files (for example `agents/openai.yaml`, `references/canvas-authoring.md` and `references/design-craft.md`).

It sits in Frontend & Design, covering MCP servers and React components. It works with Model Context Protocol, Figma, Vite and TypeScript. The repository describes itself as: A canvas design tool your agent can edit. Arrange structured UI nodes on the canvas; connect Claude, Codex, Cursor, or opencode over MCP. The licence is AGPL-3.0.

When your agent uses it

  • An agent must create
  • Modify a Loora design
  • Recover from MCP schema errors
  • Missing authoring tools

Example prompts

  • “/loora-design-guide”

Workflow steps

9 steps, taken from the first numbered list in SKILL.md.

  1. Check capability and orient. Confirm the required authoring tools are
  2. Protect the target. Confirm whether the user intends Main or a branch.
  3. Form a visual direction. Extract the audience, job, content hierarchy,
  4. Establish the system. Reuse existing tokens, themes, components, spacing,
  5. Build in meaningful batches. Use createPage for the Page and its initial
  6. Inspect after meaningful edits. Call getScreenshot on the affected Page
  7. Refine surgically. Fix the largest verified problem first. With vision,
  8. Verify structure. Re-read the affected tree or nodes. Confirm component
  9. Finish deliberately. If working on a branch, compare it with Main before

What it can do on your machine

Read from SKILL.md and the folder at commit 3dd5e87. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Loora Design Guide loads about 2.4k tokens when it runs, and up to ~20k if it reads all its reference files. Until then it costs about 143 tokens; SKILL.md has 1,192 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~143
When it runs · the whole SKILL.md, loaded when a task matches
~2.4k
With references · SKILL.md plus every file in references/, read only if the agent opens them
~20k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from lassejlv/loora at commit 3dd5e87, republished under its AGPL-3.0 licence (© lassejlv). 1,192 words, ~2,368 tokens.

Download SKILL.mdSave it as .claude/skills/loora-design-guide/SKILL.md (or your agent's skills folder). This skill also uses 6 other files; get the full folder from GitHub.
name
loora-design-guide
description
Build, edit, refine, troubleshoot, and review polished responsive product interfaces through the Loora MCP server and its structured Canvas schemas. Use when an agent must create or modify a Loora design, recover from MCP schema errors or missing authoring tools, turn a brief or reference into editable Canvas nodes, establish tokens or reusable components, add themes, interactions, or motion, work safely on a Loora branch, verify with or without image vision, or export implementation code without treating HTML, JSX, or CSS as the authoring model.

Loora Design Guide

Create real, editable Loora designs through the MCP tools. Treat the Canvas document—not generated code—as the source of truth. Work like a designer: understand the product, establish a system, build coherent sections, inspect the render, and refine the weak parts.

Gate the MCP surface first

Inspect the callable Loora tools before creating or mutating anything. Require:

  • createPage, insertNodes, and patchNodes for new design authoring
  • createComponent and createInstance for reusable component work
  • setTokens for tokens or themes
  • setAnimations and animateNodes for motion
  • branch lifecycle tools for branch work

If a required tool is absent, stop before createDesign or any partial mutation. createDesign creates only an empty record; it is not a successful Canvas design. Report the missing tools and the manifest/session mismatch. Do not substitute HTML, JSX, browser clicks, or repeated speculative calls.

If a callable tool exposes nested arguments such as ref, nodes, or patch as unknown, treat that as a schema-display limitation, not permission to guess. Use the schema reference below.

Load references by action

Before this actionRead firstRead as well when applicable
Select a design, target Main, create/use a branch, compare, propose, or applytool-workflows.mdworked-examples.md for branch theming or merge payloads
Call createPage, insertNodes, patchNodes, component, token, interaction, or motion toolsmcp-schema.mdworked-examples.md for a known-good envelope
Compose responsive layout, components, themes, interactions, or motioncanvas-authoring.mddesign-craft.md for new or materially restyled work
Review pixels with image visiondesign-craft.mdtool-workflows.md for screenshot limits
Verify without image visiontool-workflows.mdmcp-schema.md to verify effective fields and collection semantics

For a one-field text or spacing patch, read mcp-schema.md plus the target node; do not load every design reference.

Follow the core loop

  1. Check capability and orient. Confirm the required authoring tools are callable. Call getUsage if budget matters. Call listDesigns, select the target explicitly, then call getDesignContext. For an existing design, inspect the relevant area with readTree, readNode, or searchNodes.
  2. Protect the target. Confirm whether the user intends Main or a branch. Carry the same designId and optional draftId through every call. For a broad or speculative redesign, prefer a new branch. Never silently switch targets.
  3. Form a visual direction. Extract the audience, job, content hierarchy, mood, constraints, and required states from the request. If the prompt is underspecified, choose a coherent direction and state it briefly; do not default to a generic dashboard.
  4. Establish the system. Reuse existing tokens, themes, components, spacing, and typography. For a new design, define a small token set and reusable components before repeating them. Use semantic, human-readable node names.
  5. Build in meaningful batches. Use createPage for the Page and its initial hierarchy. Use insertNodes for later sections, patchNodes for atomic refinements, and moveNodes for source structure. Prefer flex/grid with fill and hug; reserve absolute positioning for intentional overlays or artwork.
  6. Inspect after meaningful edits. Call getScreenshot on the affected Page or node. With image vision, compare the pixels against the brief and design-craft.md. Without image vision, use the renderer result plus readTree/readNode as described in tool-workflows.md and state that pixel quality was not visually judged. A successful mutation alone is not proof.
  7. Refine surgically. Fix the largest verified problem first. With vision, use visible hierarchy, spacing, contrast, alignment, content, and responsive behavior. Without vision, fix only structural issues supported by reads or renderer metadata; do not invent pixel problems.
  8. Verify structure. Re-read the affected tree or nodes. Confirm component instances, interactions, token references, responsive overrides, and target revision. Use viewPage or viewNode for a canonical Loora link.
  9. Finish deliberately. If working on a branch, compare it with Main before proposing or applying it. Do not apply, close, or delete anything without the user's authority. Use exportCode only when the user needs a one-way implementation artifact.
Show full SKILL.md (567 more words)Show less

Preserve Canvas semantics

  • Send structured nodes and fields. Never insert HTML, JSX, Tailwind classes, arbitrary CSS, or code nodes.
  • Keep the three schema layers separate: the outer tool envelope, a NodeRef or descriptor locator, and the typed field value. Never move fields between those layers.
  • Keep IDs distinct: designId selects a design, draftId selects a branch, pageId selects a Page, componentId selects a component, a NodeRef selects an existing node, and descriptor ref is only a temporary label.
  • Use temporary descriptor ref values only inside a single create/insert payload. Save the permanent IDs returned in refs for later calls.
  • Use the exact NodeRef returned by readTree for component descendants. instancePath is meaningful; do not replace it with an empty array.
  • Edit a component source to change all instances. Patch an instance descendant only for a deliberate visual or content override. Do not structurally insert inside an instance.
  • Use Page or component state definitions plus declarative interactions for behavior. Do not simulate application state with hidden duplicate trees when a typed state and action expresses it.
  • Use tokens for repeated colors, numbers, and fonts. Use components for repeated structures with shared identity, not merely because two rectangles look similar.
  • Set mode on every container that will lay out children. A frame may flow within its parent while still defaulting its own children to absolute positioning.
  • Read before patching collection fields. Preserve existing Page/component states, interactions, animation lists, and an existing patch for the same breakpoint unless replacement is intentional.
  • Keep motion restrained and purposeful. Prefer the provided presets. Always judge the static composition before animating it.
  • Treat deleteNodes, deleteDesign, and closeBranch as destructive. Obtain explicit confirmation, then pass the required confirmed: true.

Work efficiently

  • Start with one context call, not a chain of broad reads.
  • Read only the subtree being changed. Use searchNodes to locate known copy or names.
  • Create a whole coherent section in one nested payload, but split very large pages by section so errors and refinements stay local.
  • Batch independent node patches into one patchNodes call.
  • Use getScreenshot at useful milestones, not after every field.
  • Inspect at the Page's intended desktop width and at least one narrow width when responsive behavior matters.
  • Report what changed, which target was used, whether verification was visual or structural, and any remaining uncertainty. Include the returned Loora URL when useful.

Handle failure without thrashing

  • On schema rejection, read the error path from the outer envelope inward, then compare it with mcp-schema.md and the live tool schema. Correct one coherent payload; do not retry variants at random.
  • If mutation tools disappear from the callable surface, stop. Do not create an empty design and hope later calls become available.
  • On an invalid or stale ID, re-read the nearest tree and use returned IDs.
  • On a locked node, stop and tell the user unless unlocking it is clearly part of the request.
  • On a read-only branch, inspect its status. Reopen a proposed branch only when the user wants further edits.
  • On merge revision drift, call compareBranch again. Never reuse stale revisions or guess conflict resolutions.
  • On CANVAS_UNAVAILABLE, explain that the legacy design is unsupported; do not attempt to reconstruct or overwrite it.
  • On screenshot failure, preserve the successful structured edit, inspect the reported error or skipped images, and be explicit that visual verification is incomplete.
  • On usage or plan limits, stop repeated calls and report the exact returned code and remaining work.

© lassejlv, AGPL-3.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 6 other files (references) in skills/loora-design-guide of lassejlv/loora.

  • SKILL.md
  • agents/openai.yaml
  • references/canvas-authoring.md
  • references/design-craft.md
  • references/mcp-schema.md
  • references/tool-workflows.md
  • references/worked-examples.md

Open the folder on GitHubat commit 3dd5e87

Compare with similar skills

Loora Design Guide next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Loora Design Guide compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Loora Design Guide this skilllassejlv/loora140—~2.4kAutomated safety check: PassAGPL-3.0
Igniteui Wc Figma To AppIgniteUI/igniteui-webcomponents170—~7.2kAutomated safety check: PassMIT
Web Artifacts Builderanthropics/skills180k40 repos~769Automated safety check: PassApache-2.0
Shadmin Devahaodev/shadmin174—~2.8kAutomated safety check: PassMIT
Wonder BlocksKhan/wonder-blocks163—~3.2kAutomated safety check: PassMIT
Typescript Styletjx666/vscode-mcp105—~961Automated safety check: PassCustom licence

Similar skills

  • Igniteui Wc Figma To App

    IgniteUI/igniteui-webcomponents

    Builds Ignite UI Web Components views from Figma designs, supporting Indigo.Design kits, third-party kits (Material 3, Fluent 2, shadcn/ui, Untitled UI, in-house), and plain frames.

    170 GitHub stars~7.2k tokensUpdated yesterday
    Frontend & DesignAuto-check passed
  • Web Artifacts Builder

    anthropics/skills

    Official

    Builds multi-component claude.ai HTML artifacts as a small React, TypeScript and Tailwind project, then bundles it into one shareable HTML file.

    180k GitHub starsUsed in 40 repos~769 tokens
    Frontend & DesignAuto-check passed
  • Shadmin Dev

    ahaodev/shadmin

    Apply Shadmin feature-development standards (backend Go/Gin/Ent + frontend React/TS).

    174 GitHub stars~2.8k tokensUpdated 2 days ago
    Frontend & DesignAuto-check passed
  • Wonder Blocks

    Khan/wonder-blocks

    Implements user interfaces using the Wonder Blocks (WB) design system — Khan Academy's React component library.

    163 GitHub stars~3.2k tokensUpdated yesterday
    Frontend & DesignAuto-check passed
  • Typescript Style

    tjx666/vscode-mcp

    TypeScript code style and repo conventions for VSCode MCP — inference vs explicit types, ESM import suffixes, zod schema contracts, async VSCode/Node APIs, JSON-safe IPC results, JSDoc for public…

    105 GitHub stars~961 tokensUpdated 2 mo ago
    DevelopmentAuto-check passed
  • MCP SDK Audit

    awdr74100/figwright

    Upgrade @modelcontextprotocol/server (the MCP TypeScript SDK v2) and prove the wire contract survived.

    1k GitHub stars~4.1k tokensUpdated today
    Agent WorkflowsAuto-check passed

Questions about Loora Design Guide

What does Loora Design Guide do?

Build, edit, refine, troubleshoot, and review polished responsive product interfaces through the Loora MCP server and its structured Canvas schemas. Loora Design Guide is an agent skill from lassejlv/loora. Build, edit, refine, troubleshoot, and review polished responsive product interfaces through the Loora MCP server and its structured Canvas schemas.

When should I use Loora Design Guide?

Loora Design Guide fits situations like: an agent must create; modify a Loora design; recover from MCP schema errors; missing authoring tools.

How do I install Loora Design Guide in Claude Code?

Run `npx skills add lassejlv/loora --skill loora-design-guide -a claude-code`. Or copy the skill folder (skills/loora-design-guide in lassejlv/loora) into .claude/skills/loora-design-guide in your project. Claude Code loads it when a task matches its description.

How do I install Loora Design Guide in Codex?

Run `npx skills add lassejlv/loora --skill loora-design-guide -a codex`. Or copy the skill folder (skills/loora-design-guide in lassejlv/loora) into .agents/skills/loora-design-guide in your project. Codex loads it when a task matches its description.

Can I use Loora Design Guide in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add lassejlv/loora --skill loora-design-guide -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/loora-design-guide, .gemini/skills/loora-design-guide, .github/skills/loora-design-guide and .opencode/skills/loora-design-guide in your project.

What does Loora Design Guide need to run?

SKILL.md names no scripts, command-line tools or credentials: Loora Design Guide is instructions for the agent only.

Does Loora Design Guide access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Loora Design Guide safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Loora Design Guide use?

Loora Design Guide is published under the AGPL-3.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Loora Design Guide use?

About 2.4k tokens (SKILL.md is roughly 9.5k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 17k tokens, read only when the agent opens those files.

What are the alternatives to Loora Design Guide?

Skills that share tags, products or a category with Loora Design Guide: Igniteui Wc Figma To App (IgniteUI/igniteui-webcomponents, 170 stars), Web Artifacts Builder (anthropics/skills, 180k stars), Shadmin Dev (ahaodev/shadmin, 174 stars) and Wonder Blocks (Khan/wonder-blocks, 163 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Loora Design Guide?

lassejlv (a GitHub user) maintains it in lassejlv/loora, which has 140 GitHub stars. The repository was last updated on September 27, 2026.

Source: lassejlv/loora on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.