Frad LEE FradSer

lark-* skill 组织策略 E2E 测试

对比两种 skill 发现面：decentralized（每个 skill description 自驱动）vs routed（子 skill 保留原版描述 + 上层 lark-router 统筹分发）。用真实 @skills/ lark-* 27 个 skill。模型：GLM-5.2。

文件

cases.py — 12 个重叠测试用例
select_test.py — 主测试：把 skill 目录喂给 headless claude -p，让其输出 PICK: <skill>。用独立 HOME 隔离全局 office:lark 插件。
SKILL.md + dispatch.md — 自编 lark-router 的两个文件（gist 扁平化，原本在 lark-router/ 与 lark-router/references/ 下）
REPORT.md — 完整结果与结论

	const orgId = document.cookie.match(/(?:^\|;\s*)lastActiveOrg=([^;]+)/)?.[1];

	if (orgId) {
	fetch(`https://claude.ai/api/organizations/${orgId}/overage_credit_grant`, {
	method: 'POST',
	headers: { 'Content-Type': 'application/json' },
	body: '{}'
	})
	.then(res => res.json())
	.then(console.log);

	#!/bin/bash

	# Claude Code Status Line - inspired by Starship configuration
	# Reads JSON input from stdin and displays contextual information

	input=$(cat)

	# Extract information from JSON input
	model_name=$(echo "$input" \| jq -r '.model.display_name // "Claude"')
	current_dir=$(echo "$input" \| jq -r '.workspace.current_dir // ""')

	# MARK: - Claude Code

	# Claude provider configurations - format: ["provider"]="url\|token"
	declare -A CLAUDE_PROVIDERS=(
	["provider"]="url\|token"
	["provider"]="url\|token"
	["provider"]="url\|token"
	)

	function claude() {

	- Follow the user's requirements exactly as provided.
	- Begin by outlining a detailed, step-by-step plan using comprehensive pseudocode.
	- Once the plan is confirmed, proceed to write the code.
	- Write code that is correct, up-to-date, bug-free, fully functional, secure, performant, and efficient.
	- Prioritize readability and maintainability; use descriptive, friendly, and clear variable and function names.
	- Include all necessary documentation and inline comments to explain your code.
	- Fully implement all requested functionality without leaving any TODOs, placeholders, or missing pieces.
	- Be concise and minimize any extraneous prose.
	- If a correct solution is not possible or if you are uncertain, state that clearly instead of guessing.

	/**
	* Generates a Nuclear Hexagram (互卦 / Hu Gua) from the provided original hexagram.
	*
	* The Nuclear Hexagram is a combination of two trigrams,
	* the Upper Trigram (上卦 / Shang Gua) and the Lower Trigram (下卦 / Xia Gua):
	* - The Upper Trigram is formed by extracting the third, fourth,
	* and fifth lines from the original hexagram.
	* - Conversely, the Lower Trigram is formed by extracting the second, third,
	* and fourth lines from the original hexagram.
	*

	[
	{
	"name": "75% Alice Layout by Frad",
	"author": "Frad LEE",
	"switchMount": "cherry"
	},
	[
	{
	"x": 1.45,
	"a": 7

	export PATH="/opt/homebrew/bin:$PATH" # Setup Homebrew

	eval "$(starship init zsh)" # Setup starship

	source $HOME/.config/dotflies/*.zsh # Setup custom dotfiles

	# Download Znap, if it's not there yet.
	[[ -f $HOME/.znap/zsh-snap/znap.zsh ]] \|\|
	git clone --depth 1 -- \
	https://github.com/marlonrichert/zsh-snap.git $HOME/.znap/zsh-snap

	# /bin/bash

	echo "Removing Oculus folder..."
	rm -rf ./Assets/Oculus ./Assets/Oculus.meta
	echo "Removing files named with oculus, OVR, AndroidManifest, vrapi, vrlib, vrplatlib..."
	find . -name "oculus" -exec rm -rf {} \;
	find . -name "OVR" -exec rm -rf {} \;
	find . -name "AndroidManifest" -exec rm -rf {} \;
	find . -name "vrapi" -exec rm -rf {} \;
	find . -name "vrlib" -exec rm -rf {} \;