audit-icon-pipeline
About
This skill audits icon pipelines by detecting missing glyphs, icons, and HD variants through registry comparisons. It generates structured gap reports for skills, agents, and teams across all palettes. Use it before rendering pipelines, after adding new entities, or for regular maintenance checks.
Quick Install
Claude Code
Recommendednpx skills add pjt222/agent-almanac -a claude-code/plugin add https://github.com/pjt222/agent-almanacgit clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/audit-icon-pipelineCopy and paste this command in Claude Code to install this skill
Documentation
Audit Icon Pipeline
以註冊表校之於字形映射文件、圖標目錄、清單,察缺字形、缺圖標、陳舊清單。生結構化之缺口報告,覆 skills、agents、teams。
適用時機
- 新增 skills、agents、teams 後,查是否需圖標
- 全管線渲染前,辨所缺
- 註冊表更新後,確清單一致
- 圖標管線之定期健檢
輸入
- 選擇性:實體類型之過濾——
skill、agent、team或all(預設:all) - 選擇性:所查之調色板(預設:
cyberpunk——參考調色板)
步驟
步驟一:讀註冊表
自真源之註冊表採諸實體之識別符。
- 讀
skills/_registry.yml——取所有域中之 skill ID - 讀
agents/_registry.yml——取所有 agent ID - 讀
teams/_registry.yml——取所有 team ID - 錄其數:skills、agents、teams 之總
預期: 三實體識別符之列,其數合於 total_skills、total_agents、total_teams。
失敗時: 某註冊表文件缺失時,報其路徑而略此實體類型。
步驟二:讀字形映射
自字形映射文件採所有已映射之實體識別符。
- 讀
viz/R/glyphs.R——取SKILL_GLYPHS列中所有鍵 - 讀
viz/R/agent_glyphs.R——取AGENT_GLYPHS列中所有鍵 - 讀
viz/R/team_glyphs.R——取TEAM_GLYPHS列中所有鍵
預期: 三已映射識別符之列。
失敗時: 某字形文件缺失時,報之並將該類實體全標為未映射。
步驟三:算缺之字形
以註冊表之識別符減已映射之識別符。
- 缺之 skill 字形:
registry_skill_ids - mapped_skill_ids - 缺之 agent 字形:
registry_agent_ids - mapped_agent_ids - 缺之 team 字形:
registry_team_ids - mapped_team_ids
預期: 存於註冊表而無字形函數映射之實體識別符之列。
失敗時: 減之算失敗時,驗註冊表與字形文件間之識別符格式是否合(如下劃線 vs 連字號)。
步驟四:查已渲之圖標
驗已映射字形有對應之已渲圖標文件。
- 每映射之 skill ID,查
viz/public/icons/<palette>/<domain>/<skillId>.webp - 每映射之 agent ID,查
viz/public/icons/<palette>/agents/<agentId>.webp - 每映射之 team ID,查
viz/public/icons/<palette>/teams/<teamId>.webp - 以同結構查
viz/public/icons-hd/中之 HD 變體
預期: 有字形而無已渲圖標之實體列(標準 / HD)。
失敗時: 若圖標目錄不存,則管線未曾運行——報全為缺。
步驟五:查清單之新鮮度
以清單之數比註冊表之數。
- 讀
viz/public/data/icon-manifest.json——計條目 - 讀
viz/public/data/agent-icon-manifest.json——計條目 - 讀
viz/public/data/team-icon-manifest.json——計條目 - 比之於註冊表之總
預期: 清單之數合於註冊表之數。有差則示清單陳舊。
失敗時: 清單文件不存時,先須運行數據管線(node build-data.js && node build-icon-manifest.js)。
步驟六:察孤兒圖標
走 viz/public/icons*/,標 WebP 文件之 <palette>/<domain>/<skillId> 三元組不在 icon-manifest.json 者。
- 列所有 WebP 文件:
find viz/public/icons* -name "*.webp" - 每文件,自其路徑取
<domain>/<id> - 查
<domain>/<id>於icon-manifest.json有無條目 - 收無匹之文件為孤兒——存於盤而已不被引
# Quick orphan count per palette
node -e "
const fs = require('fs');
const manifest = JSON.parse(fs.readFileSync('viz/public/data/icon-manifest.json'));
const ids = new Set(manifest.map(e => e.domain + '/' + e.id));
const orphans = require('child_process')
.execSync('find viz/public/icons -name \"*.webp\"').toString().trim().split('\n')
.filter(p => { const parts = p.split('/'); const id = parts.slice(-2).join('/').replace('.webp',''); return !ids.has(id); });
console.log('Orphans:', orphans.length);
orphans.forEach(p => console.log(' ', p));
"
預期: 孤兒為零。有孤兒示 skills 已遷至他域而未清(每遷 18 孤兒 = 9 調色板 × 2 尺寸)。
失敗時: 手刪孤兒——無清單條目,不再被提供。遷域事罕,手清可受。
步驟七:生缺口報告
生結構化之總。
- 格式為明表或列:
=== Icon Pipeline Audit === MISSING GLYPHS (no glyph function): Skills: 5 missing — [list] Agents: 2 missing — [list] Teams: 0 missing MISSING ICONS (glyph exists, no rendered WebP): Standard (512px): 3 skills, 1 agent HD (1024px): 8 skills, 3 agents, 1 team STALE MANIFESTS: icon-manifest.json: 320 entries vs 326 registry (stale) agent-icon-manifest.json: 66 entries vs 66 registry (OK) team-icon-manifest.json: 15 entries vs 15 registry (OK) - 依所發提下一步之建議
預期: 完整之缺口報告,附可行之下一步。
失敗時: 若諸查皆過而無缺口,報「管線已完全一致」為正果。
驗證
- 三註冊表皆讀成
- 三字形映射文件皆查
- 圖標目錄標準與 HD 皆掃
- 清單新鮮度已驗
- 孤兒圖標已查(盤路徑對清單)
- 缺口報告已生,附數與實體列
- 已提可行之下一步
常見陷阱
- 識別符格式失配:註冊表用連字號(
create-skill),字形映射或用下劃線鍵——比較時須標準化 - 調色板假設:僅查 cyberpunk 調色板漏調色板專之渲染缺口
- 空目錄:域目錄存而空,globbing 時算作「圖標在」——查文件存,非目錄存
- HD 未渲:HD 圖標在分離目錄樹(
icons-hd/)——勿與標準混 - 遷域後之孤兒:skill 之域變時,
build.sh於新路徑創圖標而不刪舊路徑——遷域後務行步六孤兒之查
相關技能
- create-glyph — 創此審所辨之缺字形
- enhance-glyph — 提升既有字形之質
- render-icon-pipeline — 運行全管線以生所缺圖標
GitHub Repository
Related Skills
content-collections
MetaThis skill provides a production-tested setup for Content Collections, a TypeScript-first tool that transforms Markdown/MDX files into type-safe data collections with Zod validation. Use it when building blogs, documentation sites, or content-heavy Vite + React applications to ensure type safety and automatic content validation. It covers everything from Vite plugin configuration and MDX compilation to deployment optimization and schema validation.
polymarket
MetaThis skill enables developers to build applications with the Polymarket prediction markets platform, including API integration for trading and market data. It also provides real-time data streaming via WebSocket to monitor live trades and market activity. Use it for implementing trading strategies or creating tools that process live market updates.
creating-opencode-plugins
MetaThis skill helps developers create OpenCode plugins that hook into 25+ event types like commands, files, and LSP operations. It provides the plugin structure, event API specifications, and implementation patterns for JavaScript/TypeScript modules. Use it when you need to intercept, monitor, or extend the OpenCode AI assistant's lifecycle with custom event-driven logic.
sglang
MetaSGLang is a high-performance LLM serving framework that specializes in fast, structured generation for JSON, regex, and agentic workflows using its RadixAttention prefix caching. It delivers significantly faster inference, especially for tasks with repeated prefixes, making it ideal for complex, structured outputs and multi-turn conversations. Choose SGLang over alternatives like vLLM when you need constrained decoding or are building applications with extensive prefix sharing.
