What Is the Hallmark Slop Test and How Many Gates Does It Have?

The Hallmark slop test is a built-in quality-control checklist with 57 distinct gates plus a pre-emit self-critique that automatically inspects every generated page for AI-tell patterns and enforces strict design constraints.

The Hallmark slop test is central to the Nutlope/hallmark repository, an automated page-generation system that refuses to ship anything that looks artificially produced. Every time Hallmark builds a page, it subjects the output to a rigorous checklist designed to catch subtle slop—the visual and structural tells that betray AI-generated design—and to enforce token budgets, color-band limits, and prohibited UI chrome.

What Is the Hallmark Slop Test?

The slop test is not an external plugin; it is baked into the Hallmark skill as a mandatory quality gate. Its purpose is to detect the faint patterns that make a composition feel machine-made and to ensure the result adheres to the macrostructure, theme, and genre selected for the run.

According to the repository’s README.md, Hallmark “runs fifty-seven slop-test gates plus a pre-emit self-critique” before it ever emits final HTML or CSS. The skills/hallmark/references/slop-test.md file then enumerates those gates in detail, describing each check that must pass unchanged. In practical terms, the slop test acts as a final auditor that can veto a page and force a full revision.

How Many Gates Are in the Hallmark Slop Test?

The slop test contains 57 distinct gates. A 58th check—the pre-emit self-critique—runs alongside them, bringing the total checklist to 58 individual evaluations.

The README.md explicitly documents the split: 57 gates are dedicated slop tests, while the pre-emit critique provides a final holistic review. Meanwhile, skills/hallmark/references/slop-test.md refers to the full battery as “the 58 gates that fire unchanged,” confirming that the pre-emit critique is counted as part of the unchanged checklist even though it functions slightly differently from the other 57. Together, these two sources make the count unambiguous.

How the Slop Test Works During a Hallmark Run

The execution flow inside the Hallmark skill is straightforward and unforgiving. As described in skills/hallmark/SKILL.md, the slop test is invoked at step 7 of the generation pipeline. The engine follows this cycle:

1. Build the page according to the chosen macrostructure, theme, and genre.
2. Run the 58-check slop-test checklist.
3. Fail the run immediately if any gate returns “no.”

When a failure occurs, Hallmark loops back, revises the output, and re-runs the entire checklist. It will not finalize the page until every one of the 57 slop-test gates returns “pass” and the pre-emit self-critique clears.

Running the Hallmark Slop Test Locally

You can trigger the slop test manually using the audit verb. This is useful when you want to inspect an existing file rather than generate a new page.

# Run the Hallmark skill audit against a target file

hallmark audit path/to/page.html

Successful output lists each gate sequentially:

# Hallmark · pre-emit critique: P5 H4 E5 S4 R5 V5

# … generated HTML/CSS …

✅ Slop-test gate 01 … passed
✅ Slop-test gate 02 … passed
…
✅ Slop-test gate 57 … passed

If a gate fails, the CLI halts, reports the failing check, and Hallmark revises the markup automatically until the audit returns a clean bill of health.

Key Files That Define and Exercise the Slop Test

Several files in the Nutlope/hallmark repository govern exactly how the slop test behaves and how it is validated:

• README.md — Introduces the slop-test concept and documents the official gate count of 57 plus the pre-emit self-critique.
• skills/hallmark/references/slop-test.md — Contains the definitive list of all slop-test gates and their detailed definitions.
• skills/hallmark/SKILL.md — Shows where the slop test is invoked at step 7 and explains how the pre-emit self-critique integrates with the rest of the gates.
• site/_tests/custom/README.md — Demonstrates the slop test in action against a custom-theme test page, providing a concrete example of gate evaluation in practice.

Summary

• The Hallmark slop test is an internal quality-control checklist that intercepts AI-tell patterns and enforces design constraints.
• It comprises 57 gates plus one pre-emit self-critique, for a total of 58 checks.
• If any gate fails, Hallmark automatically revises the page and re-runs the checklist.
• The audit command (hallmark audit) lets developers invoke the test manually against any HTML file.
• Gate definitions live in skills/hallmark/references/slop-test.md, and the orchestration logic is documented in skills/hallmark/SKILL.md.

Frequently Asked Questions

How many gates does the Hallmark slop test have?

The Hallmark slop test has 57 gates. When you include the pre-emit self-critique, the total number of checks in the quality-control run is 58.

What happens if a slop-test gate fails?

If any gate returns “no,” Hallmark fails the run immediately. The engine then revises the generated output and re-runs the full slop-test checklist from the beginning, repeating the loop until every gate passes.

Where are the slop-test gate definitions stored?

The authoritative definitions live in skills/hallmark/references/slop-test.md inside the Nutlope/hallmark repository. This file lists each gate and describes what it inspects.

How do I run the slop test on an existing page?

Use the audit CLI verb: hallmark audit path/to/page.html. This executes the full 58-check slop-test suite against the specified file and prints per-gate results in the terminal.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}