Deft
Ask AI

Search

Search skills, authors and topics

fatihaydost / brand-identity-skill

brand-identity

Brand identity sets from one brief - logo, typography and colour palette built together, measured (contrast, colour blindness, glyph coverage, font licences, small-size logo renders) and compared side by side as identity cards, shown on the user's own site, then a brand guidelines kit (PDF, logo files, tokens) for the chosen set. Use for a new brand or any part of one (logo, fonts, colours, or a mix), for refreshing an identity, or for critiquing one, even if the user only says "branding".

View sourceVisualTypographyColorSystems

Install

npx skills add https://github.com/fatihaydost/brand-identity-skill --skill brand-identity

Critique

Critique mode: "what do you think of our identity / logo / fonts / colours". Output = gates, short findings per component, the three measurable fixes that matter most. No overall score: the evidence supports no weighting. Critique the work, never the person; name what works before what doesn't. Tags: [strong] [limited] [practice].

Flow

  1. Context: sector, competitors, position, markets, languages. Read the site first; ask ≤ 3 questions; write guesses under "Assumptions".
  2. Collect: brand.py critique --site URL and/or --logo f.svg --fonts … --palette p.json --langs …. Values read from a site are approximate and labelled measured; confirm colour roles with the user before judging (site-preview.md).
  3. First impression (two seconds): what would a stranger say this brand is, what do you remember?
  4. Gates, then findings per component, then top three.

Gates (measurable or irreversible)

GatePass whenBasis
Contrasttext roles and card text ≥ 4.5:1 (large 3:1) in every shipped mode; UI parts ≥ 3:1WCAG 2.2 [strong]
Glyphsevery brand language's base exemplars in every facefont audit
Licenceeach face's licence known and covers logo, web, appfont audit
Logo filevector, no <text>/<image>, works in one colour, 16 px render existslogo.md
Confusionprimary not "the same colour" as a direct competitor's (color.md)thresholds [practice]
Local meaningno national flag as emblem ground; other readings reported as risk6ter [strong]

Unchecked gates are marked "not checked: reason", never passed.

Findings (two to four lines per component; evidence for each)

  • Logo: distinctive vs the competitor shelf; does it identify without explaining; one colour, reversed, 16 px; construction (near-equal radii, clogged joins, false holes); category clichés from ai-defaults.md §B.
  • Type: role split (voice vs reading); class against the competitor type map; weights, tnum, I/l/1; default-list hits; licence cost.
  • Palette: category legibility vs differentiation, ownability, system (roles, scales, drift: many near duplicates), producibility (dark ground, print), fit of lightness and chroma to the personality.
  • Cohesion: one mechanism or three unrelated picks? Do the components sit at the same axis positions? Is there one loud thing or several? "Feels modern" is not evidence; "primary L 0.62, C 0.19, saturated mid-light, reads energetic (Labrecque & Milne 2012, [limited])" is.

Top three

Order: failed gates (unreadable first, then confusion, licence, glyphs), then the biggest weakness against the brief, cheapest change first among equals (tokens before logo). Each fix states:

  • What: "CTA fill #4FA3E0 → #1F6FB2 (primary-700)".
  • Measured effect: "white label 2.6:1 → 5.1:1".
  • Why: one sentence tied to the brief or a finding.
  • Cost: tokens · site CSS · reprint · logo redraw. Never "improve contrast". Re-run the audit after a fix; quote before/after. A clean audit means nothing is measurably broken, not that the identity is good.

Template (plain words; numbers go to the audit file)

markdown
## Your identity: <Brand>**First impression:** <1–2 sentences>**Checks:** easy to read ✓/✗ · fonts cover your languages ✓/✗ · font licences OK ✓/✗ · logo works small and inone colour ✓/✗ · not mistaken for a competitor ✓/✗ (— not checked, why)**Logo:** … **Type:** … **Colours:** … **Together:** …**Worth keeping:** <what and why>**Three changes, most important first:** 1. … 2. … 3. …**Assumed:** <competitors, market> · **Next:** "See these fixes on your site?" / "Want refresh directions?"

Refresh = Create with that component in refresh state.