io.github.Nutlope/inspo

Measured conclusion: Deep-tested. L2 outcome: tools_ok.

Real production websites for coding agents: palettes, fonts, fold autopsies and reference JSX.

Add to your agent

Copy this into your MCP client config:

{
  "mcpServers": {
    "inspo": {
      "url": "https://inspomcp.dev/api/mcp",
      "transport": "streamable-http"
    }
  }
}

Measured tools (15)

• search_screens read-only

  Search the archive of real website screenshots. Returns screens with palette, fonts, components and image URLs.

  Parameters:

  • mode — string — light or dark
  • vibe — string — Mood / vibe - 'calm', 'loud', 'luxe', 'technical', 'soft', etc.
  • color — string — Color word - 'warm', 'cool', 'monochrome', 'neon', 'earthy'
  • limit — integer
  • query — string — Natural-language description of what you're looking for. e.g. 'minimalist editorial agency portfolio'
  • style — string — Filter to one visual style (e.g. 'minimalism', 'editorial')
  • detail — string — 'standard' (default): northstar, palette, fonts, axes, macro and tags per row, plus the fold-by-fold autopsy on the top 3. 'full': the autopsy on every row (~300 tokens each; only if you will read them all). 'concise': northstar, palette, fonts. Text profile defaults to concise.
  • device — string — 'mobile' restricts to sites with a 375px capture pair and inlines the mobile thumbnail. Default: desktop.
  • industry — string — Filter to one industry (e.g. 'saas', 'fintech', 'portfolio')
  • pageType — string — Filter to a page kind: 'landing' (default homepages), 'pricing', 'features', 'auth', 'about', 'blog', 'changelog', 'docs', 'other'
  • accentHue — string — Accent temperature: 'warm' (10-60°), 'cool' (200-300°), 'neutral' (unsaturated), 'chromatic-other' (greens, magentas, everything else).
  • maxTokens — integer — Approximate token ceiling for this response. Trims lower-ranked results and thumbnails; the top result and all URLs survive.
  • paperBand — string — Surface lightness, measured from the capture: 'dark' (<30% L), 'mid' (30-85%), 'light' (>85%). More reliable than `mode`, which is tagged and disagrees with the measurement on ~15% of rows.
  • displayClass — string — Construction of the display face: 'grotesk-sans', 'geometric-sans', 'roman-serif', 'mono', 'display-condensed-bold', etc.
  • macrostructure — string — Filter to a macrostructure (e.g. 'bento-grid')

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• get_screen read-only

  Fetch the full record for one screen by slug - every viewport variant, palette, fonts, tech stack, designer credit.

  Parameters:

  • slug (required) — string — Screen slug, e.g. 'atelier-mira'

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• get_design_system read-only

  The full DESIGN.md for one screen - real fonts, frequency-ranked palette, CSS variables, detected tech, colour roles, type ramp. Thin captured tokens are supplemented by a live fetch of the source; `live=false` skips that.

  Parameters:

  • slug (required) — string — Screen slug, e.g. 'linear-app'. Use search_screens or find_similar first to discover slugs.
  • live — boolean — If true (default), supplement thin captured tokens by fetching the source URL via study(). Set false to skip the network round-trip.

  Hints: read-only yes · destructive no · idempotent yes · open-world yes

• find_similar read-only

  A screen's nearest design neighbours - ranked by design-similarity embeddings where available, else by macrostructure / industry / style overlap. One result per site.

  Parameters:

  • slug (required) — string
  • limit — integer
  • detail — string — 'standard' (default): northstar, palette, fonts, axes, macro and tags per row, plus the fold-by-fold autopsy on the top 3. 'full': the autopsy on every row (~300 tokens each; only if you will read them all). 'concise': northstar, palette, fonts. Text profile defaults to concise.
  • sameSite — boolean — Include other pages of the same site (default false: neighbours are other sites).
  • maxTokens — integer — Approximate token ceiling for this response. Trims lower-ranked results and thumbnails; the top result and all URLs survive.

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• compare read-only

  Compare 2-4 captured sites across palette, typefaces, macrostructure, mode, style tags, type/spacing/radius scales and container width. Returns per-site breakdowns plus what they share - triangulates a house style from a few references.

  Parameters:

  • slugs (required) — array — 2-4 screen slugs to compare, e.g. ['linear-app','stripe-com','vercel-com']. Discover slugs with search_screens / find_similar.

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• find_by_color read-only

  Real sites whose extracted palette sits near a hex colour (OKLAB distance, so 'close' means the same colour family, not just the same hue). Use when a brief names a brand colour.

  Parameters:

  • hex (required) — string — Target colour as a hex string ('#c7402f', 'c7402f', '#fff'). Case-insensitive. 3- or 6-digit.
  • limit — integer
  • detail — string — 'standard' (default): northstar, palette, fonts, axes, macro and tags per row, plus the fold-by-fold autopsy on the top 3. 'full': the autopsy on every row (~300 tokens each; only if you will read them all). 'concise': northstar, palette, fonts. Text profile defaults to concise.
  • maxTokens — integer — Approximate token ceiling for this response. Trims lower-ranked results and thumbnails; the top result and all URLs survive.
  • tolerance — number — Max OKLAB distance to count as a match. Defaults to 0.15 ('same family'). 0.05 ≈ 'near-identical', 0.30 ≈ 'in the same hue zip code'.

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• find_examples_for_macrostructure read-only

  Real sites exemplifying one of the 19 named macrostructures. Call it at the macrostructure-pick step. Accepts slugs ('bento-grid') or display names ('Bento Grid').

  Parameters:

  • name (required) — string — Macrostructure name. e.g. 'bento-grid', 'Bento Grid', 'specimen', 'Marquee Hero'.
  • limit — integer
  • detail — string — 'standard' (default): northstar, palette, fonts, axes, macro and tags per row, plus the fold-by-fold autopsy on the top 3. 'full': the autopsy on every row (~300 tokens each; only if you will read them all). 'concise': northstar, palette, fonts. Text profile defaults to concise.
  • device — string — 'mobile' restricts to sites with a 375px capture pair and inlines the mobile thumbnail. Default: desktop.
  • maxTokens — integer — Approximate token ceiling for this response. Trims lower-ranked results and thumbnails; the top result and all URLs survive.

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• list_collections read-only

  List every editor-curated issue. Use this for thematic browsing - e.g. all 'Editorial Layouts' or 'Dark Product Pages'.

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• get_filters read-only

  Every value the filters accept: styles, industries, macrostructures, modes, vibes, colors, page types, component types. Zero input. Call it when unsure - an out-of-vocabulary value errors.

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• get_site_pages read-only

  A site's captured pages as an ordered flow (landing → pricing → features → ...) with titles, northstars and thumbnails - how a real product sequences its pages. No arguments returns the directory of flow-capable sites. Accepts a site or screen slug.

  Parameters:

  • limit — integer — Directory mode only: max sites to list.
  • siteSlug — string — Site slug (e.g. 'linear-app') or any screen slug from that site. Omit to list flow-capable sites instead.
  • maxTokens — integer — Approximate token ceiling for this response. Trims lower-ranked results and thumbnails; the top result and all URLs survive.

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• find_components read-only

  Real sites featuring a specific UI component (hero, pricing, features, cta, nav, footer, testimonial, logo-cloud, faq, stat), with a per-element crop where one exists. For copy-pasteable code use get_reference_jsx.

  Parameters:

  • type (required) — string — Which component type to find
  • mode — string
  • vibe — string — Mood - 'calm', 'loud', 'luxe', etc.
  • color — string — Color word - 'warm', 'cool', 'monochrome', etc.
  • limit — integer
  • style — string — Filter the parent site's style
  • device — string — 'mobile' restricts to sites with a 375px capture pair and inlines the mobile thumbnail. Default: desktop.
  • industry — string
  • pageType — string — Only components from this page kind (e.g. 'pricing')
  • maxTokens — integer — Approximate token ceiling for this response. Trims lower-ranked results and thumbnails; the top result and all URLs survive.
  • macrostructure — string

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• get_collection read-only

  Fetch one issue by slug, including the editor's blurb and ordered screen list.

  Parameters:

  • slug (required) — string — Collection slug, e.g. 'editorial-layouts'
  • detail — string — 'standard' (default): northstar, palette, fonts, axes, macro and tags per row, plus the fold-by-fold autopsy on the top 3. 'full': the autopsy on every row (~300 tokens each; only if you will read them all). 'concise': northstar, palette, fonts. Text profile defaults to concise.
  • maxTokens — integer — Approximate token ceiling for this response. Trims lower-ranked results and thumbnails; the top result and all URLs survive.

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• find_reference_components read-only

  The canonical reference components as an index: id, label, macrostructure and a note for each hero / pricing / cta / nav / footer archetype. Pick one, then get_reference_jsx(type, id) for its source.

  Parameters:

  • type — string — Component category (hero, pricing, cta, …). Omit to scan all types.
  • macro — string — Substring match on the component's macro field - e.g. 'Marquee' matches 'Marquee Hero'.

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• recommend read-only

  Orchestrator - start here for a brief. One call returns a macrostructure pick, 5 real exemplars, matching reference components (source via get_reference_jsx), a palette suggestion, and an `evidence` packet measuring the genre's paper band / display class / accent hue. Pass a macrostructure to skip the pick step. No LLM call.

  Parameters:

  • brief (required) — string — The user's design brief in plain English. E.g. 'calm meditation app', 'dark dev-tool that helps teams ship faster'.
  • mode — string — Optional: 'light' or 'dark'.
  • vibe — string — Optional vibe filter.
  • color — string — Optional colour word.
  • detail — string — 'standard' (default): northstar, palette, fonts, axes, macro and tags per row, plus the fold-by-fold autopsy on the top 3. 'full': the autopsy on every row (~300 tokens each; only if you will read them all). 'concise': northstar, palette, fonts. Text profile defaults to concise.
  • device — string — 'mobile' restricts to sites with a 375px capture pair and inlines the mobile thumbnail. Default: desktop.
  • pageType — string — Optional: 'landing', 'pricing', 'features', etc.
  • maxTokens — integer — Approximate token ceiling for this response. Trims lower-ranked results and thumbnails; the top result and all URLs survive.
  • macrostructure — string — Optional: a macrostructure already picked. Skips the pick step.

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• get_reference_jsx read-only

  The full canonical JSX source for one reference component, including the header stamp naming its macrostructure / theme / states. Copy-pasteable; tweak tokens to the target brand.

  Parameters:

  • type (required) — string — Component category
  • id (required) — string — Reference id within that type (e.g. 'marquee', 'three-card').

  Hints: read-only yes · destructive no · idempotent yes · open-world no

Repository

Measured by MCPYP probing. Untested fields are omitted, never estimated. Listing is always free; MCPYP never proxies calls or stores user tokens.