Copy and paste this message to your agent to have this skill automatically installed for you.

Install the apple-hig skill for me from https://github.com/justinwetch/HIGAgentSkills.git

An agent skill containing 156 distilled Apple Human Interface Guidelines reference files, structured for accurate, context-efficient design guidance across Apple platforms.

The Apple HIG is authoritative and comprehensive, but its full text is too large and too prose-heavy to load directly into an agent's context. This skill distills each HIG topic into a compact, information-dense reference file that preserves specific rules, measurements, API names, and platform distinctions while removing redundant explanatory prose.

The result is a corpus that fits within a practical context budget and gives the agent precise, citable answers rather than paraphrased recollections.

This branch updates the skill against the Apple Human Interface Guidelines corpus captured on 2026-06-09.

Highlights:

• 156 distilled HIG topic files
• 16 tier-1 foundation files loaded on every invocation
• Restored Design Principles coverage as a tier-1, easy-triggered reference
• Updated routing index with 1,057 trigger keywords
• Current-source review completed for every inventoried source page
• Full-corpus audit and continuity sampling recorded under sources/apple-hig-2026-06-09/verification/

Total: 156 files, 1,057 trigger keywords, ~140k rough tokens in the full corpus.

Platforms covered: iOS, iPadOS, macOS, tvOS, visionOS, watchOS.

SKILL.md defines a 7-step loading protocol:

1. Parse the request for platform(s), component names, and framework references
2. Load all 16 tier-1 foundation files (always, every invocation)
3. Load the matching designing-for-[platform] file based on detected platform
4. Keyword-scan the request against routing-index.md and load all tier-3 matches
5. Expand one hop through each loaded file's related: frontmatter
6. Load tier-4 files only on direct keyword match
7. Answer citing exact values from loaded files

The routing-index.md file maps 1,057 trigger keywords to their corresponding reference files across all four tiers.

SKILL.md               Agent entry point and loading protocol
README.md              Skill overview and release/install notes
routing-index.md       Keyword-to-file routing map (auto-generated)
distilled/             156 runtime reference files, one per HIG topic
  *.md

Each distilled/*.md file includes YAML frontmatter with topic, tier, platforms, category, triggers, and related fields. The agent reads these fields to navigate the corpus without loading every file.

Production testing should use the script-generated runtime zip, not the full repository.

Build the Claude-safe package with:

python scripts\package_runtime_zip.py

This writes release/apple-hig.zip. The archive contains exactly one top-level folder named apple-hig/ with:

apple-hig/
  SKILL.md
  README.md
  routing-index.md
  distilled/
    *.md

The packaging script writes forward-slash paths only, omits directory metadata, validates archive paths, and excludes hidden files, macOS resource forks, sources/, scripts/, process.md, .gitignore, raw Apple captures, rendered pages, caches, and git metadata. Those files are useful for repo review and future maintenance, but they are not required for an installed skill and should not be loaded by production agents.

To test locally, unzip or install the apple-hig/ folder into your agent skills directory, then ask Apple platform design questions that should trigger tier-1 foundations, platform files, and component-specific files.

The floor cost per invocation is approximately 33,600 rough tokens (tier-1 files plus the routing index). Typical queries load 20 to 30 files and land between roughly 34,000 and 55,000 tokens, within a practical budget on a 200k context window.

Each reference file was distilled from the corresponding Apple HIG source page, targeting a 75% reduction in word count. The compression preserved:

• Specific measurements, sizes, and spacing values
• API names, class names, and framework identifiers
• Platform distinctions and per-platform behavioral rules
• Do/do not rules stated as direct directives

Removed: introductory framing, change history, repetitive examples, and prose that restated rules already expressed concisely elsewhere.

Each compression pass was followed by an evaluation step to confirm that no substantive rules or specifications were lost, only extraneous prose. Files were only accepted when the distilled version could answer the same design questions as the source.

Use the runtime zip for production testing. After extraction, the installed skill directory must be named apple-hig so the folder name matches the name: apple-hig metadata in SKILL.md.

The skill triggers when a request involves Apple platform design, HIG components, specific measurements, Apple framework integration, or general interface design work that should use Apple's Design Principles as grounding.

Content distilled from the Apple Human Interface Guidelines, available at developer.apple.com/design/human-interface-guidelines. This repository is an independent reference tool and is not affiliated with or endorsed by Apple Inc.