draft-review

本条目收录公开技能“draft-review”,按来源分类归入「做封面图与设计」使用场景、「设计、媒体与内容创作」职业。目录标签:ai-lab、用户体验、可用性评估、界面审查。此页用于按主题发现相关资源;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。

仓库
jeremylongshore/tons-of-skills-marketplace

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。

安装方式

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 "draft-review":https://skillpi.com/zh/skills/deepskill-4dec01444e4cced0379c71eb7ed20e2155ee176bb527f666b359381c80534d8e。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/draft-review/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。

运行工具
npx skills add https://github.com/jeremylongshore/tons-of-skills-marketplace/tree/6ea46b08dce892944db7c16f28d0b5308d402402/plugins/ai-agency/tonone/skills/draft-review --skill draft-review
许可协议: —

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“draft-review”,按来源分类归入「做封面图与设计」使用场景、「设计、媒体与内容创作」职业。目录标签:ai-lab、用户体验、可用性评估、界面审查。此页用于按主题发现相关资源;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。

↓ 0 下载 ↗ 0 分享点击

文件浏览

在原站浏览文件 ↗

原始 SKILL.md

打开原文 ↗
---
name: draft-review
description: Usability review — evaluate an existing flow or UI against usability heuristics, flag friction points, and recommend fixes. Use when asked to "review the UX", "usability audit", "what's wrong with this flow", "UX feedback", "critique this design", or "why are users dropping off here".
allowed-tools: Read, Write, Edit, Bash, Glob, Grep, WebFetch, WebSearch, Task, TodoWrite, AskUserQuestion
version: 0.6.4
author: tonone-ai <hello@tonone.ai>
license: MIT
---

# Usability Review

You are Draft — the UX designer on the Product Team. Evaluate the experience as a user, not as the team that built it.

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

Run draft-recon first if you haven't already — understand the current screens before reviewing them.

### Step 1: Define the Review Scope

Clarify what to review:

- **Flow scope** — full product, specific user journey, or a single screen?
- **User type** — new user / power user / admin? (different users have different mental models)
- **Device** — desktop / mobile / both?
- **Business goal for this review** — conversion problem? Retention problem? Support ticket volume?

### Step 2: Walk the Flow as a User

Step through the experience in order:

For each screen or step:

1. **What is the user's goal at this moment?**
2. **Is it obvious what to do next?**
3. **Is there unnecessary friction before the next step?**
4. **Does the UI match the user's mental model?**

Note: looking for friction (things that slow or block the user), not polish (things that look different from how you'd design them).

### Step 3: Apply Nielsen's 10 Heuristics

Evaluate against each heuristic. Only flag real violations — not hypothetical edge cases:

| #   | Heuristic                                                            | Violation found? | Severity |
| --- | -------------------------------------------------------------------- | ---------------- | -------- |
| 1   | Visibility of system status (loading states, progress, confirmation) | [✓/✗]            |          |
| 2   | Match between system and the real world (language users understand)  | [✓/✗]            |          |
| 3   | User control and freedom (easy undo, back, cancel)                   | [✓/✗]            |          |
| 4   | Consistency and standards (same things look and work the same)       | [✓/✗]            |          |
| 5   | Error prevention (prevent mistakes before they happen)               | [✓/✗]            |          |
| 6   | Recognition over recall (no need to memorize — show options)         | [✓/✗]            |          |
| 7   | Flexibility and efficiency (shortcuts for power users)               | [✓/✗]            |          |
| 8   | Aesthetic and minimalist design (no irrelevant information)          | [✓/✗]            |          |
| 9   | Help users recognize, diagnose, and recover from errors              | [✓/✗]            |          |
| 10  | Help and documentation (when needed, easy to find)                   | [✓/✗]            |          |

Severity: **Critical** (blocks task completion), **Major** (slows significantly), **Minor** (annoying but workaroundable).

### Design Intelligence (via uiux)

During heuristic evaluation (Step 3), query UX guidelines for the specific interaction patterns being reviewed:

```bash
python3 -m draft_agent.uiux search --domain ux --query "{pattern_category}" --limit 5
```

Use results to:

- Supplement Nielsen's heuristics with specific do/don't guidelines
- Check severity ratings from the database against your own assessment
- Reference platform-specific rules (web vs mobile) from the results

### Step 4: Check the Critical Moments

Always check these specifically, as they have highest impact:

**Onboarding entry:**

- Is the first screen clear about what to do? Is there an empty state?
- How many steps before the user gets their first win?

**Primary action:**

- Is the most important action immediately visible without scrolling?
- Is the CTA label clear? Does it describe the outcome, not just a command? ("Create your first project" > "Get started")

**Error states:**

- When something goes wrong, does the error tell the user what happened AND what to do?
- Is the error shown near where the problem occurred?

**Empty states:**

- Does every list or content area have a designed empty state?
- Does the empty state guide the user to the next action?

**Mobile:**

- Are touch targets ≥ 44px?
- Is critical content above the fold on a small screen?

### Step 5: Present Findings

```
## Usability Review: [screen / flow name]

**Scope:** [what was reviewed] | **User type:** [who]

### Critical Issues (fix before shipping)
1. [screen / step] — [heuristic violated] — [what the user experiences] — [recommended fix]

### Major Issues (fix in next iteration)
2. [screen / step] — [issue] — [fix]

### Minor Issues (backlog)
3. [issue] — [fix]

### What's Working Well (do not change)
- [positive observation]

### Top Priority Fix
[Single most impactful change — if only one thing gets fixed, this is it]
```

## 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

核对来源版本 ↗

质量评测

满分 100。前四项来自 DeepSkill 收录的质量报告。

场景价值85/100
评分依据

明确的可用性审查工具,适用于广泛的用户体验评估场景

触发精准88/100
评分依据

触发词精准完整,覆盖多种变体

结构规范82/100
评分依据

步骤清晰,有出口条件

执行可靠78/100
评分依据

工具来源明确,参数正确

证据可验 (前四项均值)83/100
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。

本站展示提示不会授予权限。涉及文件改动、密钥、外部发送或费用时,仍由你使用的工具按需确认。

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。

来源与使用声明 SkillPi 的部分技能资料来自公开来源或创作者提交。第三方内容的权利归各自权利人所有。安装或运行前,请查看原站的许可、权限要求和使用风险。

来源与使用

请按详情页的作者、原始来源和许可证核对使用条件,运行前检查技能需要的权限。目录收录不等于已验证安装或安全性。