hallmark/site/_tests/verbs/study/notes.md

study verb · what fired and why

Input

A described screenshot (input-description.md) of a Pentagram-style design studio's portfolio homepage. Italic-serif studio mark, numbered project list, warm cream paper, single warm-orange accent on the active row, Single-Line Colophon footer.

What study does

"Extract the DNA — macrostructure, archetypes, type-pairing role, colour anchor, rhythm — and produce a diagnosis report, then optionally rebuild the user's content using the extracted DNA. Never copies pixels. Never claims to identify exact fonts. Refuses obvious template-marketplace screenshots."

Five-step pipeline (per SKILL.md § hallmark study):

1. Refuse-or-proceed check — public reference, not a paid template → proceed.
2. Vision pass — read the image into the structured-fields schema in study.md. Output ten fields.
3. Diagnosis report — return the named macrostructure + archetypes + anti-patterns to NOT carry over. See diagnosis.md.
4. Confirmation question — ask before building.
5. Build — produce output.html + output.css with the user's actual six projects, stamped studied: yes.

What loaded

• references/study.md — the protocol, schema, and refusal heuristics.
• references/study-examples.md — the three worked examples (Pentagram-style portfolio, Klim-specimen, Rauno-personal). The Pentagram example matched our screenshot most closely; the skill imitated the diagnosis structure.
• references/macrostructures.md — to name the page-shape (Specimen).
• references/component-cookbook.md — for archetype names (H4 Margin-Number Display, Ft1 Single-Line Colophon).
• references/typography.md — for the type-role vocabulary (italic-display + sans-body) and free / paid candidate pairings.
• references/color.md — for the warm-orange-at-low-chroma colour-anchor.
• references/microinteractions.md — for the hover-duration anti-pattern callout (the screenshot's 800 ms is too slow; should be 150–200 ms).

What study did not do

• Did not name the exact font. Visual font ID is unreliable; the skill named the role (italic editorial serif at high optical size) and proposed two real candidates (Fraunces free / Tiempos Headline paid).
• Did not pixel-clone. The output uses Coelho Studio's six actual projects, not "Aperture's" five-or-whatever projects. The bones come from the screenshot; the words and projects come from the user.
• Did not copy the screenshot's hover duration. The screenshot used 800 ms; the rebuild uses 180 ms. The skill flagged this in the diagnosis as anti-pattern-to-not-carry-over.
• Did not copy the screenshot's footer text size. The screenshot used 11 px; the rebuild uses 13 px. Below 14 px is below the body floor (per typography.md); the skill flagged this and bumped it.
• Did not copy any imagery, photography, or proprietary content.

Stamp

The output's first CSS comment block:

/* Hallmark · macrostructure: Specimen · H4 hero knobs: number-column=margin, alignment=left-bias
 * theme: Specimen · accent: warm-orange ~3% · enrichment: none (typography only)
 * studied: yes · DNA-source: user reference (Aperture-style portfolio) · paid-fonts: not confirmed (free pairing applied)
 */

The studied: yes flag tells future audit runs to be lenient on "Specimen-fall-through" (gate 23) — the user explicitly chose this DNA — but stricter on "did the page actually use the extracted DNA?" The paid-fonts: not confirmed flag documents the substitution to the user.

Why this is the canonical study test

The screenshot maps neatly to study-examples.md Example A (Pentagram-style editorial portfolio). It exercises:

• The refusal heuristics (public reference, not a paid template → proceed).
• The 10-field structured schema.
• The font-role-not-font-ID discipline.
• The "anti-patterns the screenshot has — do NOT carry over" output.
• The confirmation question.
• The studied: yes stamp.

If the input had been a paid Squarespace-template marketplace screenshot, the verb would have refused. If the input had been ambiguous (an unknown agency page or a personal portfolio with no clear ownership), the verb would have asked before extracting. Both branches are documented in study.md; this test exercises the proceed-with-public-reference branch.