Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
git clone --depth 1 https://github.com/fledgeling-co/fledgeling-pluginsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/plugins/fledgeling-co/fledgeling-plugins/ux-craft)<a href="https://agentmods.dev/plugins/fledgeling-co/fledgeling-plugins/ux-craft"><img src="https://agentmods.dev/badge/plugins/fledgeling-co/fledgeling-plugins/ux-craft/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/plugins/fledgeling-co/fledgeling-plugins/ux-craft"><img src="https://agentmods.dev/badge/plugins/fledgeling-co/fledgeling-plugins/ux-craft.svg" alt="Reviewed on agentmods" width="80" height="20"></a>Grade A, and why
ux-craft scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 10 lines — stays where its author put it; the contents beside it link to each section on GitHub.
{
"name": "ux-craft",
"version": "2.5.3",
"description": "Book-grounded UX engine for building, mocking and reviewing web and mobile UIs, forms, flows and emails — the UX brain that pairs with design-craft:design-craft's visual hands. Three modes (Build / Review / Advise) over eleven references, and a deterministic gate that refuses what prose only asks: stdlib-only ux-lint.py runs static over HTML/JSX/CSS and probes a rendered page through Obscura, with every finding naming what you did, what the user silently gets, and the fix. Only exit 0 is a pass — a run that examined zero files refuses rather than reporting a clean sheet, and every run prints a never-empty not-checked list, because a check that cannot measure must say so instead of reporting zero. v2 resolves the accessibility floor against the standards themselves: 24x24 CSS px is WCAG 2.2 SC 2.5.8 at AA and is the only target-size number a WCAG failure may cite, 44x44 CSS px is SC 2.5.5 at AAA, and Apple's 44pt and Android's 48dp are craft targets in density-independent units that are not WCAG numbers at all. Twelve non-negotiables each carry the silent symptom that appears when they break, the state grid is a counted artifact rather than a rule that was read, and the destructive-action gate table, the probe-honesty rule and the model-family evidence taxonomy are available to every model rather than quarantined behind a model check. Carries a Known limits section, an evidence file rating the replication status of every behavioural law it cites (including the ones it argues against), the measured blind spots of the only sanctioned browser, and an autonomous-mode carve-out so a pipeline run is not told both to ask and not to ask. Build mode names the category-default journey and varies it on one axis, with a persistent project ledger so two onboardings are not the same three-step wizard. Persuade surfaces load a separate one-offer/one-audience/one-action conversion reference.",
"author": {
"name": What it installs
The manifest is a name and a version. 1 skill travel with it, and installing the plugin installs all of them — 256 tokens a session between them. Each is measured on its own page, and each can be installed alone.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago Changed a33c65a03fd0
- 1mo ago First seen · 10 lines scan A 0272f4407f8a
ux-craft is a plugin published in the GitHub repository fledgeling-co/fledgeling-plugins (2 stars, last pushed 4d ago), licensed MIT. Its token cost is not measured: this kind of file is read by the harness, not the model. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-04.
Other plugins, from other repositories
frontend-design
Frontend design skill for UI/UX implementation.
frontend-design
Astro, frontend design, and web design guidelines skills.
aio-visual-diff
Verify AI-built UI against design via measurement-driven diff — extracts computed CSS and bounding boxes through Chrome DevTools MCP, diffs against Figma reference or a frozen baseline, and feeds numerical deltas back to the agent. No screenshot eyeballing.
liquid-glass
Apple-style liquid glass: frosted, refracting surfaces for web UI, with the rules that keep them from reading as gray boxes.
shadcn
UI component and design system framework. Search registries, install components as source code, and audit your project.
layout
Serve and enforce your design system in any agent. Bundles the Layout MCP server (design tokens, component specs, compliance checks) and the layout-md skill for writing on-brand UI code.