# How Hallmark Scores Philosophy, Hierarchy, Execution, Specificity, Restraint, and Variety

> Discover how Hallmark uses its six-axis self-critique system to score Philosophy Hierarchy Execution Specificity Restraint and Variety ensuring high-quality artifact generation through automatic revisions when scores dip below 3

- Repository: [Hassan El Mghari/hallmark](https://github.com/Nutlope/hallmark)
- Tags: deep-dive
- Published: 2026-07-31

---

**Hallmark evaluates every generated artifact before emission using a six-axis pre-emit self-critique system that scores Philosophy, Hierarchy, Execution, Specificity, Restraint, and Variety on a 1–5 scale, triggering automatic revision if any axis falls below 3.**

The Hallmark framework by Nutlope implements a rigorous quality gate for generative design artifacts. According to the source code in [`skills/hallmark/SKILL.md`](https://github.com/Nutlope/hallmark/blob/main/skills/hallmark/SKILL.md) and [`skills/hallmark/references/slop-test.md`](https://github.com/Nutlope/hallmark/blob/main/skills/hallmark/references/slop-test.md), every output undergoes a mandatory pre-emit critique across six independent design axes to prevent "slop" and ensure high-fidelity results.

## The Six Design Axes Explained

Hallmark’s scoring rubric evaluates each artifact against six specific criteria defined in [`skills/hallmark/references/slop-test.md`](https://github.com/Nutlope/hallmark/blob/main/skills/hallmark/references/slop-test.md). Each axis receives a score from 1 (poor) to 5 (excellent).

### Philosophy

**Philosophy** measures the existence of a clear *why*—a design position or narrative that guides the page rather than a generic layout. A score of **5** indicates an explicit, compelling purpose, while **3** suggests some rationale but vague execution. A **1** means no discernible intent exists. According to the logic in [`SKILL.md`](https://github.com/Nutlope/hallmark/blob/main/SKILL.md) lines 46–48, scores below 3 trigger a revision to redesign the brief or add a conceptual hook.

### Hierarchy

**Hierarchy** assesses whether a reader can grasp primary, secondary, and tertiary information within **2 seconds**. A **5** requires distinct visual hierarchy with clear typographic scale, while **3** indicates partial hierarchy (primary clear but secondary ambiguous). A **1** represents a flat composition where all elements appear equal weight. Revision triggers include adjusting typographic scale, color weight, or spacing to emphasize hierarchy.

### Execution

**Execution** evaluates fidelity to the detailed design spec, including rule weight, accent footprint, text-wrap, focus rings, and contrast. **5** represents pixel-perfect implementation with no sloppiness, **3** indicates minor inconsistencies, and **1** reveals obvious sloppy details like mis-aligned borders or missing accessibility hooks. Scores below 3 require fixing low-level CSS/HTML errors and adding missing accessibility hooks.

### Specificity

**Specificity** checks alignment with the *specific brief*—the output should look like a bespoke solution, not a generic template. **5** means tailored to the specific brief, **3** indicates some generic patterns remain, and **1** looks like a one-size-fits-all page. When scores fall below 3, Hallmark replaces default macrostructures or components with brief-specific variants.

### Restraint

**Restraint** measures the removal of any element that does not earn its place, targeting extraneous decoration and "padding-for-padding’s-sake." **5** represents a lean, purposeful composition, **3** allows some unnecessary flourishes, and **1** indicates cluttered, over-decorated output. Revision involves pruning superfluous UI parts and simplifying color or shape usage.

### Variety

**Variety** calculates structural distance from previous Hallmark outputs in the same project, examining macrostructure, nav/footer archetypes, and theme axes. **5** requires a clearly distinct fingerprint, **3** allows minor variations, and **1** means the same structure as recent builds. The system consults [`.hallmark/log.json`](https://github.com/Nutlope/hallmark/blob/main/.hallmark/log.json) to compare against previous macrostructures; scores below 3 trigger selection of a different macrostructure or theme according to diversification rules.

## How the Pre-Emit Critique Works

The scoring process operates as a mandatory gate before any artifact reaches the user.

### The 1–5 Rubric and Revision Logic

As implemented in [`skills/hallmark/SKILL.md`](https://github.com/Nutlope/hallmark/blob/main/skills/hallmark/SKILL.md), Hallmark generates a draft page or component, then runs the six-axis audit. If **any axis scores below 3**, the system loops back to the design flow, revises the problematic dimension, and re-runs the critique until all scores are ≥ 3. This ensures the final output respects the six disciplines before proceeding to the 58-gate "slop test."

### The Pre-Emit Stamp Format

After scoring, Hallmark inserts a one-line stamp comment at the very top of the artifact. The format follows the specification in [`slop-test.md`](https://github.com/Nutlope/hallmark/blob/main/slop-test.md) lines 24–25:

```css
/* Hallmark · pre‑emit critique: P5 H4 E5 S4 R5 V5 */

```

The letters correspond to **P**hilosophy, **H**ierarchy, **E**xecution, **S**pecificity, **R**estraint, and **V**ariety respectively. This stamp serves as both documentation and a machine-readable quality marker.

## Source Code Implementation

The scoring system relies on specific files within the repository:

- **[`skills/hallmark/SKILL.md`](https://github.com/Nutlope/hallmark/blob/main/skills/hallmark/SKILL.md)** – Declares the pre-emit self-critique discipline and the "< 3 triggers revision" rule (lines 46–48).
- **[`skills/hallmark/references/slop-test.md`](https://github.com/Nutlope/hallmark/blob/main/skills/hallmark/references/slop-test.md)** – Provides detailed axis definitions, scoring criteria, and stamp format (lines 15–25).
- **[`site/css/tokens.css`](https://github.com/Nutlope/hallmark/blob/main/site/css/tokens.css)** – Holds token definitions that the Execution and Restraint axes check against (e.g., `var(--color-accent)`).
- **[`.hallmark/log.json`](https://github.com/Nutlope/hallmark/blob/main/.hallmark/log.json)** – Generated at runtime to record previous macrostructures and themes; consulted by the Variety axis to enforce diversification.

The automated scoring routine follows this pattern:

```javascript
function scoreArtifact(artifact) {
  const scores = {
    P: assessPhilosophy(artifact),
    H: assessHierarchy(artifact),
    E: assessExecution(artifact),
    S: assessSpecificity(artifact),
    R: assessRestraint(artifact),
    V: assessVariety(artifact),
  };
  // Insert stamp at top of artifact
  const stamp = `/* Hallmark · pre‑emit critique: P${scores.P} H${scores.H} E${scores.E} S${scores.S} R${scores.R} V${scores.V} */\n`;
  return stamp + artifact;
}

```

The concrete implementations of the `assessX` helpers examine layout hierarchy, token usage, accessibility hooks, and compare the current macrostructure against the project’s log to determine the Variety score.

## Summary

- Hallmark uses a **six-axis pre-emit critique** (Philosophy, Hierarchy, Execution, Specificity, Restraint, Variety) to evaluate every generated artifact.
- Each axis scores **1–5**, with **< 3 triggering automatic revision** until all thresholds are met.
- Scores are embedded as a **stamp comment** at the top of each file using the format `P# H# E# S# R# V#`.

- The system references **[`slop-test.md`](https://github.com/Nutlope/hallmark/blob/main/slop-test.md)** for rubric definitions and **[`SKILL.md`](https://github.com/Nutlope/hallmark/blob/main/SKILL.md)** for the revision logic.
- **Variety** enforcement relies on [`.hallmark/log.json`](https://github.com/Nutlope/hallmark/blob/main/.hallmark/log.json) to prevent repetitive macrostructures across project iterations.

## Frequently Asked Questions

### What happens if one axis scores below 3 but others are perfect?

If any single axis scores below 3, Hallmark **must revise the artifact** regardless of other scores. As defined in [`skills/hallmark/SKILL.md`](https://github.com/Nutlope/hallmark/blob/main/skills/hallmark/SKILL.md), the system loops back to the design flow, addresses the specific weakness (such as adjusting hierarchy or pruning decoration), and re-runs the full six-axis critique until all scores reach 3 or higher.

### How does Hallmark calculate the Variety score?

The Variety axis examines the **structural fingerprint** of the current output against previous builds recorded in [`.hallmark/log.json`](https://github.com/Nutlope/hallmark/blob/main/.hallmark/log.json). It checks for differentiation in macrostructure, navigation/footer archetypes, and theme axes. A score of 5 requires a clearly distinct approach, while repetition of recent patterns results in a 1, triggering the selection of alternative macrostructures or themes.

### Where does the scoring data appear in the final output?

Hallmark inserts a **machine-readable stamp comment** at the very first line of every generated artifact (CSS, HTML, etc.) following the format `/* Hallmark · pre‑emit critique: P# H# E# S# R# V# */`. Additionally, scores appear in the preview block shown to users, allowing verification before code shipment.

### What distinguishes the Execution axis from Specificity?

**Execution** measures technical fidelity and polish—pixel-perfect alignment, correct rule weights, accessible focus rings, and proper contrast. **Specificity** evaluates conceptual alignment with the brief—whether the solution looks bespoke or generic. A design could execute perfectly (high Execution) yet still look like a template (low Specificity), or be highly specific but technically sloppy, triggering different revision paths.