AI coding agents are now fast enough to build a whole web page in minutes — but the pages often look the same: the same purple gradients, the same rounded cards, the same generic layout. Developers have a name for that look: AI slop. Hallmark is an open-source skill built to stop exactly that. Created by Hassan El Mghari (@nutlope, Together AI), it sits alongside a coding agent — Claude Code, Cursor, Codex or OpenCode — and checks the agent's design work against a fixed set of quality rules before anything ships.
Hallmark is not a design tool you open; it is a skill an agent reads. It ships as a SKILL.md file plus reference material and Node scripts, including a browser linter called `sloplint` that inspects the rendered page. The heart of it is 57 deterministic quality gates spread across 6 axes, so the checks are repeatable rather than a matter of taste: the same page fails or passes the same way every time. Version 2.0.0, released on 21 October 2026, rewrote the skill and added 50 built-in themes. It is MIT-licensed, and installs with one command: `npx skills add nutlope/hallmark`.
Key facts
| Item | Details |
|---|---|
| Repository | nutlope/hallmark (GitHub) |
| Creator | Hassan El Mghari (@nutlope, Together AI) |
| What it is | Anti-AI-slop design skill for coding agents |
| Works with | Claude Code, Cursor, Codex, OpenCode |
| Quality gates | 57 deterministic gates across 6 axes |
| Themes | 50, in v2.0.0 (21 October 2026) |
| Linter | `sloplint` browser linter (Node scripts included) |
| Licence | MIT |
| Install | `npx skills add nutlope/hallmark` |
| Repo created | 27 April 2026 |
| Public launch | 19 May 2026 |
| GitHub stars | 58,597 as of 21 November 2026 (RepoTrending); SkillHub counted 58,682 the same day |
The growth is recent and steep: the repo had about 12,200 stars in mid-July 2026 and roughly 46,500 by 3 November. Most of the current attention belongs to the v2.0.0 rewrite of 21 October 2026 — that is the version people are starring now. Star counts measure interest, not production use: a star means a developer found the idea worth bookmarking, not that the skill is running in their workflow.
What this means
For students and working developers, Hallmark is a sign of how agent-assisted coding is maturing. The first wave of tools was about generating code quickly; the new wave is about making the output good enough to ship. A deterministic checklist — 57 gates that run the same way every time — is a very different thing from asking a model "does this look good?" It turns taste into a test, and tests are something a team can agree on, automate, and keep improving.
It also shows how the "agent skills" idea is spreading. Skills like this one are small, portable packages of instructions and checks that any compatible agent can pick up. Hallmark's own rise — from 12,200 stars in July to nearly 59,000 by late November — is part of a wider rush in 2026 to give coding agents better judgement, not just more speed.
A note on verification
The star figures in this post are snapshots taken on the dates shown, from RepoTrending and SkillHub, and will have moved since. The feature details (57 gates, 6 axes, 50 themes, `sloplint`, the v2.0.0 date) are as stated in the project's own materials and the verified research row. No claim here has been checked against the live repository beyond the 21 November 2026 snapshot.
Source: nutlope/hallmark repository and star trackers (RepoTrending, SkillHub), via verified research. Last checked: 21 November 2026.
FAQs
What is Hallmark?
Hallmark is an open-source design skill (nutlope/hallmark) that helps AI coding agents produce web pages that do not look generic or "AI-made". It gives the agent a SKILL.md file, reference material and Node scripts, including a browser linter called sloplint, plus 57 deterministic quality gates across 6 axes.
Which coding agents does Hallmark work with?
It is built for Claude Code, Cursor, Codex and OpenCode. It installs with the command `npx skills add nutlope/hallmark` and is MIT-licensed.
How popular is Hallmark on GitHub?
It had 58,597 stars as of 21 November 2026 (RepoTrending; SkillHub counted 58,682 the same day), up from about 12,200 in mid-July 2026. Remember that stars measure interest, not production use.
What changed in version 2.0.0?
Released on 21 October 2026, v2.0.0 rewrote the skill and added 50 built-in themes. The current wave of stars is largely for this rewritten version.
Comments (0)
Share your question or experience about this syllabus. Your comment will appear after review.
No comments yet — be the first to share your thoughts!