LP

leonardobetti/pd-portfolio-review

开发工具

本页由 AI 翻译,技术细节请以原文为准。 查看原文

27 stars 质量 71 趋势 26MIT

招聘评审对你设计作品集的看法 粘贴一个网址。获取一份评分报告,列出优点、错误、警示信号以及最重要的修复。

概览

一项AI技能,能像忙碌的招聘评审员一样审查产品设计师作品集(以及用户体验/UI作品集)。该技能基于UXfolio的2026年UX作品集现状研究。 它还融合了Karl Koch的招聘经理-bar技能(karlkoch.me)原则:将招聘经理视为用户,仔细标记证据,并保持修正小巧且可检查。该技能获得MIT许可;参见技能/PD-portfolio-review/evals/LICENSE-hiring-manager-bar。 与UXfolio无关。评分权重是该技能自有的。 这个仓库是:可安装技能放在技能/下,营销着陆网站放在应用/网页(Next.js)。用npx技能add安装技能;从同一个仓库运行或部署网站。网站上:pd-portfolio-review.vercel.app。 1. 你要求审核并提供作品集网址(可选:目标职位、密码)。2. 技能像审核员一样打开网站:首页、工作列表、案例研究、关于、联系方式。

概述(AI 翻译)

一项AI技能,能像忙碌的招聘评审员一样审查产品设计师作品集(以及用户体验/UI作品集)。该技能基于UXfolio的2026年UX作品集现状研究。

它还融合了Karl Koch的招聘经理-律师技能(karlkoch.me)原则:将招聘经理视为用户,仔细标记证据,并保持修正小巧且可检查。该技能获得MIT许可;参见skills/pd-portfolio-review/evals/LICENSE-hiring-manager-bar。

与UXfolio无关。评分权重是该技能自有的。

该仓库是一个monorepo:可安装技能位于skills/,营销登陆网站位于apps/web(Next.js)。使用npx skills add安装该技能;在同一仓库运行或部署网站。实时站点:pd-portfolio-review.vercel.app。

路径 它是什么
skills/pd-portfolio-review 可安装的代理技能、参考与评估
apps/web 公共登陆页:安装命令、工作原理、评分、Claude.ai 上传路径、常见问题解答

它是如何运作的

  1. 你要求审核并提供作品集网址(可选:目标职位,密码)。

本页由 AI 翻译,技术细节请以原文为准。

README

以下为原文 README 节选,完整内容请查看原文。

pd-portfolio-review

An AI skill that reviews product designer portfolios (and UX / UI portfolios) the way a busy hiring reviewer would. It is informed by UXfolio’s State of UX Portfolios 2026 research.

It also incorporates principles from Karl Koch’s hiring-manager-bar skill (karlkoch.me): treating the hiring manager as the user, tagging evidence carefully, and keeping fixes small and checkable. That skill is MIT-licensed; see skills/pd-portfolio-review/evals/LICENSE-hiring-manager-bar.

Not affiliated with UXfolio. The scoring weights are this skill’s own.

This repository is a monorepo: the installable skill lives under skills/, and the marketing landing site lives under apps/web (Next.js). Install the skill with npx skills add; run or deploy the website from the same repo. Live site: pd-portfolio-review.vercel.app.

Path What it is
skills/pd-portfolio-review The installable agent skill, references, and evals
apps/web Public landing page: install command, how it works, score, Claude.ai upload path, FAQ

How it works

  1. You ask for a review and give a portfolio URL (optional: target role, password).
  2. The skill opens the site like a reviewer: home, work list, case studies, About, contact. It notes what is clear in the first ~30 seconds, then reads deeper.
  3. You get a scored report: strengths, problems, a red-flag check, open checks, questions only you can answer, and at most five fixes in priority order, each with a done-check.

Evidence on every finding:

  • Rendered: seen on the live page
  • Source: in HTML / metadata / text, not confirmed visually
  • Unverified: likely but unconfirmed (listed under Open checks; costs no points)

Score

Out of 100. Maths stay visible in the report.

Area Points
Access and format (opens, loads, no download, website) 15
First scan (clear who, what, where to click) 15
Project choice and cards (relevance, 3 or 4, card context) 15
Thinking and decisions 20
Outcomes and ownership 15
Voice, AI use, visual quality 10
About, consistency, NDA handling 10

Overrides: broken or locked link with no password scores 0 and stop. Each present red flag (Rendered or Source only) takes 3 points off the area total.

Verdict bands: 80+ ready to send · 60 to 79 send after fixes · 40 to 59 needs work first · under 40 rebuild.

Criteria

Checks follow UXfolio 2026 chapters (first 30 seconds, format and access, finding the right project, inside a case study, NDA work, AI use), including the eight ranked red flags. Figures are cited as “UXfolio 2026, ch. X.Y”, not rounded or invented.

Fairness rules built into the skill:

在 GitHub 查看完整 README

推荐工具

换一个关键词,或者移除筛选条件。

安装

npx skillfish add leonardobetti/pd-portfolio-review
新手?先看 mcpskill.org 教程 →