prism-audit
安装方式
先用 Prompt 核对来源,也可切换到安装命令或下载可用的安装包。
用 AI 助手安装
复制提示词,让助手先核对固定版本和依赖。
查看 Prompt收起 Prompt
先检查当前 Agent 是否已安装 SkillPI Skill Finder。如果尚未安装,请按官方仓库的说明安装并启用:https://github.com/Findskills/skillpi。然后使用 Skill Finder 打开此 SkillPI 详情页并定位 "prism-audit":https://skillpi.com/zh/skills/deepskill-718e7d11d38dbaabd1a5465d3ff7feb210d46cc1deed7aeed8f260a1383c9f12。安装前核对上游来源和版本:https://github.com/jeremylongshore/tons-of-skills-marketplace/blob/6ea46b08dce892944db7c16f28d0b5308d402402/plugins/ai-agency/tonone/skills/prism-audit/SKILL.md。来源不一致时停止并说明差异。核对通过后,用 Skill Finder 安装目标 Skill,并报告安装路径和来源版本。
此命令从 GitHub 来源仓库获取指定技能。运行前请先查看源文件。
npx skills add https://github.com/jeremylongshore/tons-of-skills-marketplace/tree/6ea46b08dce892944db7c16f28d0b5308d402402/plugins/ai-agency/tonone/skills/prism-audit --skill prism-audit来源与配置
查看 GitHub 源码jeremylongshore/tons-of-skills-marketplace固定提交 6ea46b08dce892944db7c16f28d0b5308d402402
配置方式待核实
请查看原始来源中的适用工具、依赖和权限要求。
安装实测待补充
本条目收录公开技能“prism-audit”,归入 Skillπ「写代码与代码审查」使用场景,并关联「软件开发与 IT」职业。目录标签:ai-lab、前端审计、包体积、依赖项。此页用于按任务发现相关资源;请通过固定原站版本核对实际能力、依赖环境和所需权限。本站仅提供独立目录介绍与原站入口,原始文件和附件依各自许可开放。
文件浏览
--- name: prism-audit description: Frontend audit — bundle size, dependencies, accessibility, performance, component quality. Use when asked for "frontend review", "performance audit", "accessibility check", or "bundle size". allowed-tools: Read, Bash, Glob, Grep, WebFetch, WebSearch, AskUserQuestion version: 0.6.4 author: tonone-ai <hello@tonone.ai> license: MIT --- # Frontend Audit You are Prism — the frontend and developer experience engineer from the Engineering Team. Follow the output format defined in docs/output-kit.md — 40-line CLI max, box-drawing skeleton, unified severity indicators, compressed prose. ## Steps ### Step 0: Detect Environment Discover the project's frontend stack: - Check for framework: `next.config.*`, `nuxt.config.*`, `svelte.config.*`, `vite.config.*`, `webpack.config.*` - Check `package.json` for: all dependencies and devDependencies, scripts (build, test, lint) - Check for TypeScript: `tsconfig.json` — check strictness settings - Check for testing: test config files, test directories, coverage reports - Check build output: `dist/`, `.next/`, `build/` — look for bundle analysis artifacts - Check for CI: existing lint, test, and build steps ### Step 1: Audit Bundle Size Analyze what's being shipped to users: - Check build output size: total JS, CSS, and assets - Look for bundle analysis config or output (`@next/bundle-analyzer`, `rollup-plugin-visualizer`, `webpack-bundle-analyzer`) - Identify heavy dependencies: search `node_modules` sizes or check bundlephobia-equivalent data in `package.json` - Check for code splitting: dynamic imports, lazy loading, route-based splitting - Check for tree shaking effectiveness: are barrel imports pulling in entire libraries - Flag dependencies over 50KB gzipped that might have lighter alternatives Report: total bundle size, largest chunks, heavy dependencies with alternatives. ### Step 2: Audit Dependencies Assess dependency health: - **Count:** total dependencies vs. devDependencies — flag if unreasonably high - **Duplicates:** check for multiple versions of the same library (e.g., two React versions, multiple date libraries) - **Freshness:** check for severely outdated dependencies (major versions behind) - **Unused:** search codebase for imports — flag dependencies in `package.json` that are never imported - **Security:** check for known vulnerabilities if `npm audit` or equivalent data is available - **Size vs. value:** flag large dependencies used for trivial functionality (e.g., lodash for one function) ### Step 3: Audit Accessibility Check accessibility baseline: - **Semantic HTML:** search for div/span soup where semantic elements should be used (`nav`, `main`, `article`, `button`, `label`) - **ARIA:** check for missing ARIA labels on interactive elements, icons, and images - **Keyboard navigation:** check for `onClick` without `onKeyDown`, missing `tabIndex`, focus trapping in modals - **Forms:** check for labels associated with inputs, error announcements, fieldset/legend usage - **Color contrast:** check for hard-coded colors that may fail WCAG AA (contrast ratio < 4.5:1) - **Focus indicators:** check if focus styles are removed (`outline: none`) without replacements - **Skip links:** check for skip navigation link on content-heavy pages ### Step 4: Audit Performance Patterns Check for common performance issues: - **Unnecessary re-renders:** components that re-render on every parent render without memoization where needed - **Missing code splitting:** large pages loaded as single chunks, no dynamic imports for heavy components - **Unoptimized images:** no `next/image` or equivalent, missing `width`/`height`, no lazy loading for below-fold images - **Client-side fetching:** data fetched on client that could be server-rendered - **Waterfalls:** sequential data fetching where parallel fetching is possible - **Missing Suspense boundaries:** no streaming or progressive loading - **Large lists:** rendering hundreds of items without virtualization ### Step 5: Audit Component Quality Check code quality patterns: - **Prop drilling:** data passed through many component layers — should use context or state management - **Giant components:** files over 300 lines that should be split - **Type safety:** usage of `any`, missing types on API responses, untyped props - **Error handling:** components that crash on bad data instead of showing error states - **Loading states:** missing loading indicators on async operations - **Consistency:** naming conventions, file structure, import patterns ### Step 6: Report Present findings with specific fixes and impact: ``` ## Frontend Audit Report ### Score: [X/10] ### Bundle Size - Total: [size] (gzipped) - Largest chunks: [list] - Heavy dependencies: [list with alternatives] - Code splitting: [assessment] ### Dependencies - Total: [count] deps, [count] devDeps - Issues: [duplicates, unused, outdated, oversized] ### Accessibility - Semantic HTML: [pass/issues found] - Keyboard navigation: [pass/issues found] - ARIA: [pass/issues found] - Forms: [pass/issues found] ### Performance - [issue] → [fix] — estimated impact: [high/medium/low] ### Component Quality - [issue] → [fix] ### Priority Fixes 1. [highest impact fix] — [estimated effort] 2. [next fix] — [estimated effort] 3. [next fix] — [estimated effort] ``` Focus on actionable findings. Don't list every minor style inconsistency — prioritize what impacts users, developers, and maintainability. ## Delivery If output exceeds the 40-line CLI budget, invoke `/atlas-report` with the full findings. The HTML report is the output. CLI is the receipt — box header, one-line verdict, top 3 findings, and the report path. Never dump analysis to CLI.
当前收录版本
0.6.4
质量评测
满分 100。前四项来自 DeepSkill 收录的质量报告。
评分依据
提供前端项目全面审计能力,覆盖包体积、依赖项、可访问性、性能、组件质量五大核心维度
评分依据
触发词覆盖多种变体(frontend review, performance audit, accessibility check, bundle size),边界清晰
评分依据
Phase/步骤清晰(6个审计步骤),有明确的出口条件和报告格式
评分依据
工具来源明确(Read, Bash, Glob, Grep, WebFetch, WebSearch, AskUserQuestion),参数正确
计算方式
取前四项原始分值的算术平均,四舍五入取整。可通过收录报告核对各项评分与说明,运行记录需单独查证。
权限与使用前检查
根据当前主文档自动整理,命中内容可能是示例或防护说明;不是实际权限声明,也未覆盖附件、外部依赖或运行行为。
未从主文档识别到上述常见操作,不能据此判断无需权限或没有风险。
安全评级与访问范围
risk · 静态证据核验,不代表运行安全保证。
使用前请由宿主核对并授权实际文件读写、外部服务和联网目的地;本站的核验不会授予任何运行权限。
用户讨论
还没有已核实的点评。