moai-designer| brand context → logo type → constraints → generation delegation → alignment verification (generation execution:moai-media)
Overview
A logo is a brand's smallest and longest-used asset. Unlike hero images, it must be legible at every size from a 16px favicon to a billboard, and must be adaptable to color, grayscale, and monochrome. That is why a logo is approached not as a raster visual but as vector/symbol.
Before generation, this skill reads the project's brand context (tokens and voice), selects a logo type, then converts those values into recraft vector prompt constraints and hands them off to generation.
What it does not do
- Does not execute image generation, select models, or disclose costs —
moai-media:media-higgsfield-coreand delegates to the contract (§5 Generation Delegation). The model is of the recraft_v4_1 vector family, but live parameter lookup, costs, and polling are all handled by the delegate. - Does not define design tokens — the source of truth for tokens is the project's design system. This skill only reads and uses them; it does not create new ones.
- Does not ask the user directly — if a slot is empty, it returns a blocker and the orchestrator checks it.
Trigger keywords
logo, brand mark, wordmark, lettermark, monogram, symbol, logotype, pictogram, app icon, favicon, logo concept, logo idea, logo redesign, emblem, combination mark, abstract mark
Reference files
| File | When |
|---|---|
references/logo-types.md |
Classification of the 7 logo types and a guide for choosing "which type to pick" |
references/logo-craft.md |
recraft_v4_1 prompt craft · designer terminology glossary · R1/R3 application |
references/logo-principles.md |
5 principles of logos · scalability · variation system · alignment verification checklist |
Workflow
Step 1 — Read brand context
Before writing the generation prompt , find and read the project's source of truth. Priority:
- Design system handoff directory (token CSS, screen design source, brand rules document)
DESIGN.mdor an equivalent brand identity document- Token files (DTCG / CSS variables / shadcn config)
- Brand voice document (voice, prohibitions, mascot usage rules)
If none exists, proceed without it, but do not invent colors without token evidence. State in the result that there is no evidence.
Step 2 — Select logo type
Determine the logo type from the user request and brand personality (references/logo-types.mdsee the selection guide):
| Type | When it fits |
|---|---|
| Wordmark | Name is short, recognizable, or the name itself is the differentiator |
| Lettermark | Name is long or the acronym is already familiar |
| Symbol/Pictogram | When a universal shape connects with the brand |
| Abstract | A distinctive geometric mark without a concrete form |
| Mascot | The character is the brand personality |
| Combination | Both a symbol and a wordmark are needed |
| Emblem | Traditional, official, institutional image |
If the user has not specified a type and it cannot be determined from brand context alone, raise 2–3 candidates as a blocker so the orchestrator can confirm. To explore at once, you can also generate multiple concepts in parallel as a combination count .
Step 3 — Convert to constraints
Convert the read brand context into recraft prompt constraints. This conversion is the core output of this skill.
| Brand item | Prompt constraint |
|---|---|
| Palette | Specify exact colors in hex, prohibit other hues |
| Typography | For wordmark/lettermark, font character (serif/sans/script/mono) |
| Form language | Angular/rounded, geometric/organic |
| Mood | Calm/bold/warm/playful |
| Prohibition rules | Not negation but positive descriptionconvert to (→ Core R1) |
Caution when converting prohibition rules: If you put "don't use gradients" as is, the model focuses on gradients. Describe what you want, like "flat solid fills only."
Logos often have text at their core, so R3 (literal text in quotes + font specification) is especially important — the wordmark "GLOW" is "the word 'GLOW' in a geometric sans-serif" like.
Step 4 — Assemble the recraft prompt
Assemble the recraft_v4_1 prompt with the selected logo type and constraints (references/logo-craft.md):
- global-to-local structurestack core concept → form silhouette → layout → color → mood in that order.
- Designer terminologyuse:
contained mark,horizontal lockup,lettermark,negative space cutout,monoline style,serif letterforms. Everyday words ("pretty logo") blur the intent. - Avoid texture/material language — describe a vector logo as "no texture, no gradients, consistent stroke."
- Color as prompt + hex together — "green and charcoal palette, deep green (#1b6e4a) with neutral charcoal (#2b2b2b)".
If in the logo exploration stage, count: 4generating multiple concepts in parallel is effective.
Step 5 — Delegate generation
The assembled prompt and constraints moai-media:media-higgsfield-core hands it over to the contract. Model (recraft_v4_1) selection and live parameter lookup (models_explore) and cost preflight (get_cost) and polling (job_status) are all handled on that side.
If moai-media is not installed: output the completed prompt as text and guide the user to generate directly on the Higgsfield web (https://higgsfield.ai)에서 with the recraft model. Do not substitute with stock images, raster icons, or CSS gradients.
Step 6 — Consistency verification
Check the received logo against the criteria unique to logos(references/logo-principles.md):
- Scalability: does the form survive when scaled down (favicon 16px)
- Simplicity: are there unnecessary details or textures
- Variation tolerance: is it recognizable when converted to a silhouette (solid color)
- Palette consistency: does it deviate from the brand palette
- Text: if it is a wordmark/lettermark, is the spelling and lettering accurate
If it deviates, strengthen the constraints and regenerate. Do not let a deviating logo pass as "mostly right" — a logo is a kind of criterion that does not allow approximations.
Output format
## 로고 결과
- 로고 타입: [워드마크 | 심볼 | 콤비네이션 | ...] · 컨셉 수: [N]
- 참조한 브랜드 컨텍스트: [읽은 파일 경로] (없으면 "근거 없음" 명시)
- 적용 제약: [팔레트 hex / 타이포 / 형태 언어 / 무드]
- 생성 위임: moai-media:media-higgsfield-core (recraft_v4_1)
- 결과 URL: [생성 결과들]
- 정합 점검: [통과 | 어긋난 항목과 조치]
- 권장 변형 산출: [컬러 / 흑백 / 단색 실루엣 / 앱 아이콘]
Precautions
- Approach the logo as a vector/symbol — do not use texture, material, photographic, or "realistic" language.
- Do not invent colors or fonts without token evidence. If there is no evidence, state that there is none.
- Convert prohibitions into positive statements (R1).
- Wrap wordmark/lettermark text in quotation marks and specify the font together (R3).
- Do not excessively include slogans or additional text inside the logo — it breaks when made small.
- Generation cost and model are not decided by this skill — follow the delegation target's prior notification procedure as is.
- AI-generated results are not vector masters. Since recraft output may be pixel/raster, specify in the results that the final logo master (SVG) is a separate vectorization/tracing post-processing step. This skill handles up to concept generation.
Related skills
| Skill | Aspect |
|---|---|
moai-media:media-higgsfield-core |
Delegation: generation execution, cost, model |
moai-media:media-higgsfield-image |
Adjacent: recraft prompt craft SSOT |
moai-designer:design-brand-visual |
Sibling: hero, OG, mockup visuals |
moai-designer:design-brand-system |
Preceding: brand design system definition |
moai-designer:design-slop-check |
Following: AI slop check |
Source
- Higgsfield MCP — generation execution path.
- Recraft — vector/logo/icon prompt craft (primary). Global-to-local structure, designer terminology dictionary.
- Higgsfield Skills (official agent documentation) (MIT) — reference for logo usage classification.
- Generation contract, cost, model selection:
moai-media:media-higgsfield-corelive contract. - The sources for logo type classification and the 5 principles are stated in
references/logo-types.md,references/logo-principles.mdeach file.