Next AI Draw.io
draw.io diagrams your AI agent draws on a live browser canvas that you can also edit by hand
Apple Human Interface Guidelines for Agents
如果你是作者,可把这段代码贴到 README。徽章会链接回本页。
[](https://mcp.popnx.com/en/skills/justinwetch-higagentskills)在同一工作中把 AI 连到服务的 MCP
draw.io diagrams your AI agent draws on a live browser canvas that you can also edit by hand
Create, inspect, validate, and save editable 3D building scenes with Pascal's MCP server.
Verified React chart generation: select, validate, repair, render, and inspect charts through MCP.
同类工作,排序与 Skill 列表相同
| 本项 HIGAgentSkills | impeccable | diagram-design | frontend-slides | guizang-ppt-skill | |
|---|---|---|---|---|---|
| 作者 | justinwetch | pbakaus | cathrynlavery | zarazhangrui | op7418 |
| Skill 数 | 1 | 1 | 1 | 1 | 1 |
| GitHub 星标 | 836 | 111,852 | 67,796 | 32,458 | 29,958 |
| 最近代码更新 | 2026年10月18日 | 2026年11月17日 | 2026年11月13日 | 2026年6月24日 | 2026年10月30日 |
| 许可证 | MIT | Apache-2.0 | MIT | MIT | AGPL-3.0 |
justinwetch 提供的 1 个 Skill(指令集),AI 读取后照做 · 不是 MCP server
MCP = 手
把 AI 连到外部服务或数据,例如读取 Notion 或开发票。需要有程序或服务器在运行。
Skill = 手册
教 AI 怎么做某项工作的文本文件,不连接任何服务。
有些会重叠:少数 Skill 带有调用工具的脚本,少数 MCP 只提供知识。请从你想完成的工作出发。
开发者未说明 · SKILL.md 形式的 Skill 可用于支持 Agent Skills 的应用,如 Claude Code、Codex 和 Cursor
GitHub account justinwetch (person), 4 years old
★ 836 stars · code last changed 2026年10月18日
22 scripts that can run on a computer: scripts/assess_native_hig_capture.py, scripts/capture_hig_agent_evidence.py, scripts/capture_native_hig.py, scripts/check_hig_candidate.py and 18 more
Scripts that go online: scripts/capture_native_hig.py
No hooks or writes to machine settings found
Checked automatically from the repo files at commit 2206ff9 (read only, nothing was run). Not a safety guarantee: read it and decide for yourself
第一步:你用哪个 AI?
选择你用的应用,本页只显示该应用的安装方法
Uploading to claude.ai would fail: the description is over 200 characters (762). Use "Let the AI install it" instead
复制这句话,粘贴到 Claude Code 的对话中
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
重新打开 Claude Code,输入 /HIGAgentSkills,或直接提出相关任务,Claude 会自动选用
Checked against official docs 2026年11月12日 · Source: code.claude.com
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
git clone --depth 1 https://github.com/justinwetch/higagentskills.git /tmp/popnx-skill-justinwetch-higagentskills-2206ff9 && mkdir -p ~/.claude/skills && cp -R '/tmp/popnx-skill-justinwetch-higagentskills-2206ff9/.' ~/.claude/skills/higagentskillsChatGPT 桌面应用内置 Codex,Codex 从 ~/.agents/skills 文件夹读取 Skill
在 ChatGPT 桌面应用中打开 Codex,把这句话粘贴到对话中
The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
把 apple-hig 文件夹放到 ~/.agents/skills/apple-hig
~ 是你的主文件夹($HOME)
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
git clone --depth 1 https://github.com/justinwetch/higagentskills.git /tmp/popnx-skill-justinwetch-higagentskills-2206ff9 && mkdir -p ~/.agents/skills && cp -R '/tmp/popnx-skill-justinwetch-higagentskills-2206ff9/.' ~/.agents/skills/higagentskillsChecked against official docs 2026年11月15日 · Source: learn.chatgpt.com
不能使用你自己安装的 Skill,只能使用目录中插件自带的 Skill
想在 ChatGPT 中使用这个 Skill,请按上面的方法使用 ChatGPT 桌面应用
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
复制这句话,粘贴到 Codex 的对话中 · 适用于 ChatGPT 桌面应用中的 Codex、Codex CLI 和 IDE 扩展
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Codex 会自动发现新安装的 Skill;如果没有出现,请重启 Codex · 使用方法:在 ChatGPT 中输入 @ 并选择 Skill · 在 Codex CLI 或 IDE 扩展中运行 /skills,或输入 $ 加 Skill 名称
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
git clone --depth 1 https://github.com/justinwetch/higagentskills.git /tmp/popnx-skill-justinwetch-higagentskills-2206ff9 && mkdir -p ~/.agents/skills && cp -R '/tmp/popnx-skill-justinwetch-higagentskills-2206ff9/.' ~/.agents/skills/higagentskills复制这句话,粘贴到 Cursor 的对话中
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
在 Cursor 的 Agent 聊天中输入 / 并选择 HIGAgentSkills
Checked against official docs 2026年11月12日 · Source: cursor.com
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
git clone --depth 1 https://github.com/justinwetch/higagentskills.git /tmp/popnx-skill-justinwetch-higagentskills-2206ff9 && mkdir -p ~/.cursor/skills && cp -R '/tmp/popnx-skill-justinwetch-higagentskills-2206ff9/.' ~/.cursor/skills/higagentskills适用于其他能读取 Skill 的应用。这句话列出了 Claude Code、Codex 和 Cursor 的路径,AI 会按你的应用选择 · Copy this sentence into the app’s chat
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Checked against official docs 2026年11月12日 · Source: code.claude.com, learn.chatgpt.com, cursor.com
其他支持 Agent Skills 的应用(如 Gemini CLI、GitHub Copilot)使用不同文件夹,请查看该应用的指南,再把含 SKILL.md 的文件夹复制过去。
已于 2026年11月11日 对照 Claude Code、Codex 和 Cursor 官方指南核实 skill 文件夹
✓ 命令已检查 2026年11月18日
以下文字来自开发者的 repo,不是 PopNx MCP 的指令
显示前 8,000 个字符 · 在 GitHub 打开完整文件
---
name: apple-hig
description: >
Apple Human Interface Guidelines reference, updated for iPhone Duo and
OS 27. Provides authoritative platform-specific design rules, component
specifications, exact measurements, and interaction patterns for iOS,
iPadOS, macOS, tvOS, visionOS, and watchOS. Use when designing, reviewing,
auditing, or fixing any Apple platform UI, or when asked about specific HIG
components, sizing, system behaviors, frameworks (HealthKit, SiriKit, ARKit,
etc.), or platform conventions. Also use when the user is designing any
digital interface and could benefit from established design principles, even
without an explicit Apple platform context. Includes a guided /duo workflow
that adapts an existing iOS app to iPhone Duo, plus opt-in enforce mode with
independent review.
---
# Apple HIG Skill
Use the distilled HIG references in `references/hig/` through `routing-index.md`. The source snapshot and verified release measurements are in README.md.
## Loading Protocol
Release **2026-09-27**; Apple source snapshot **2026-09-12**.
For help, installation, version or capability questions, read only this entry
point and the relevant README section. Do not load the HIG corpus for setup help.
Installing or upgrading this skill includes its `/duo` slash command. After the
skill folder is in place, check whether the host already offers `/duo`. If it
does not, copy `commands/duo.md` into the host's custom slash-command or custom
prompt directory (user-level unless the user asks for one project), confirm
`/duo` is available, and tell the user if a restart is needed.
Work directly in the conversation and the user's project; do not create a browser
intake form, onboarding site, dashboard, or browser setup step.
Ordinary guidance is the default. Only an explicit request to apply enforce mode
activates its procedure. Read `references/enforce.md` first in that case and check
reviewer, Python and observation capabilities before loading the design guidance.
A leading `/duo`, `Use apple-hig in Duo mode`, or an explicit request to adapt
an existing app for iPhone Duo starts Duo mode: read `references/duo.md` and
follow its stages. Questions about the command and ordinary Duo design questions
stay ordinary guidance. Duo mode works only in the app's native code; it never
produces HTML, SVG, or web mockups.
Use the available Python 3.10+ interpreter (`python3` or `python`); do not assume
the executable name. Ordinary guidance also works without Python.
When Python is available, use the standard-library routing helper to resolve
steps 2–6: save only the exact user request to a UTF-8 text file in a temporary
directory outside the user's project, then run `<python> "<skill-root>/scripts/hig_route.py" --request-file "<request.txt>"`.
Read every file in its `files` list in bounded sections, completing any truncated
read before continuing. Do not concatenate the entire set into one tool output;
the helper output is metadata, not the guidance itself. Do not add
your own instructions to the request text or promote inferred topics to
initial matches. If the user explicitly excludes a platform/device/topic, pass
the excluded topic IDs with `--exclude-topic` (repeat for each topic, including
platform aliases matched inside an excluded phrase); record the user's reason.
Keep a platform that is also positively requested elsewhere. Do not exclude
topics merely to reduce context. Exclusions apply
to direct and related selections, never to the 16 foundations. Without Python,
apply the same steps and explicit exclusions manually.
### Step 1 — Parse Task Context
From the user's request, identify:
- **Platform(s):** ios / ipados / macos / tvos / visionos / watchos
- **Device/form factor:** iPhone Duo when named or matched by its tier-2 triggers
- **Components or patterns** mentioned (buttons, tab bars, sheets, etc.)
- **Frameworks or SDKs** referenced (HealthKit, SiriKit, ARKit, etc.)
- **Task type:** design / review / spec / audit / guidance
- **Mode:** ordinary guidance by default. Activate enforce mode only when the user explicitly asks to apply it; mentioning the mode or asking what it means does not activate it. Duo mode starts only as described above.
### Step 2 — Load Tier 1 (every design or review task)
Read all 16 files listed in `routing-index.md` under `## tier-1`:
`accessibility`, `branding`, `color`, `dark-mode`, `design-principles`,
`icons`, `images`, `inclusion`, `layout`, `materials`, `motion`,
`privacy`, `right-to-left`, `sf-symbols`, `typography`, `writing`
These apply universally. Load them before answering.
### Step 3 — Load Platform File (Tier 2)
Read `routing-index.md` → `## tier-2 platform-map`.
For each detected platform, load its `designing-for-[platform]` file.
Also match the tier-2 map's device/form-factor and API triggers using
standalone word/phrase/API-symbol matching. A Duo match loads both
`designing-for-iphone-duo` and `designing-for-ios`; Duo is an iOS form factor.
If "game" or "gaming" is mentioned, also load `designing-for-games`.
### Step 4 — Keyword Scan (Tier 3)
Read `routing-index.md` → `## tier-3 trigger-map`.
Normalize the user request to lowercase. Match trigger strings as
standalone words, phrases, or API symbols; plural, hyphenated and bare-symbol
forms count (`tab bars`, `full-screen`, `keyboardType`). Never match a trigger
merely because it appears inside an unrelated word. For example, `AR` doesn't
match "tab bar," and `AI` doesn't match "explain." Load every matching
file. Load each file at most once.
### Step 5 — Related Expansion
Freeze the initial selection set as the tier-2 and tier-3 files selected in steps 3–4. Read only that set's `related:` frontmatter.
Load the listed files once. Do not expand related lists from those newly added files.
Bind each initial topic to the exact platform/device/trigger text matched in
the user's request. An inferred
component concept is not a literal trigger match.
A topic reached through `related:` stays related even when it is useful;
never promote it to initial and expand it again. If semantic relevance calls
for an additional reference, load that specific file without related expansion.
### Step 6 — Tier 4 On-Demand
Read `routing-index.md` → `## tier-4 trigger-map (on-demand)`.
Use the same standalone word/phrase/API-symbol matching rule. Load a
tier-4 file only on direct keyword match, or if named in the initial selection
set's `related:` list. Never expand from foundations, newly related files, or
directly matched tier-4 files. These are niche — avoid loading broadly.
### Step 7 — Answer or enforce
For explicitly requested enforcement, carry out `references/enforce.md` using
the loaded source guidance. Ordinary guidance never starts that procedure.
For an ordinary guidance answer, quote the short operative source clause verbatim
with all qualifiers intact, then give only the needed application. A compact
table works for multiple states. Keep the rule and application distinct;
do not paraphrase the quotation into a stronger, broader, or narrower rule.
Before sending, delete sentences that do not answer a requested part.
Check the actor, starting state, action and resulting state; do not substitute
a similar neighboring scenario. Mention an API as the implementation of a
rule only when the loaded source supports that relationship.
Do not add unrequested adjacent rules or measurements merely because they are
loaded. Preserve the source's full scope: do not invent narrower device/state
limits when applying a rule, and include every requested state it covers.
Apply the guidance for the named platform, device, orientation, and state.
An explicit scoped exception takes precedence over a broader rule within that scope;
apply that exception only to its stated platform, device and state.
Do not invent a reconciliation for unresolved source conflicts; identify the conflict.
Apply all other compatible loaded content. If a topic isn't covered by loaded f提问、分享技巧或报告无法使用。按“有帮助”数量排序。