Agentic Awesome Skills 智能自動(dòng)分類實(shí)戰(zhàn)指南從關(guān)鍵詞打分到索引驅(qū)動(dòng)的目錄治理【免費(fèi)下載鏈接】agentic-awesome-skillsAAS Core is the local, agent-first control plane for complete catalog discovery, agent-owned selection, stack validation, and planning, backed by 2,115 agentic skills. Includes CLI, local MCP, catalog, plugins, and Workbench.項(xiàng)目地址: https://gitcode.com/gh_mirrors/an/agentic-awesome-skills導(dǎo)讀本文圍繞 agentic-awesome-skillsAAS倉(cāng)庫(kù)中docs/maintainers/smart-auto-categorization.md所定義的能力系統(tǒng)講解技能庫(kù)如何基于關(guān)鍵詞打分、優(yōu)先級(jí)規(guī)則與作用域匹配自動(dòng)消除 uncategorized 狀態(tài)并最終以skills_index.json為唯一事實(shí)來(lái)源驅(qū)動(dòng) Web 應(yīng)用分類篩選。讀完本文你將掌握自動(dòng)分類腳本的運(yùn)行與 Dry-Run 預(yù)演方法、13 個(gè)內(nèi)置類別的完整關(guān)鍵詞體系、自定義關(guān)鍵詞擴(kuò)展方式以及如何在新增技能時(shí)通過 frontmatter 聲明類別實(shí)現(xiàn)一次寫入、永久歸類。一、背景為什么要自動(dòng)分類隨著技能庫(kù)增長(zhǎng)到 2,000 個(gè) SKILL.md手動(dòng)維護(hù)類別既不現(xiàn)實(shí)也容易產(chǎn)生大量未分類條目。倉(cāng)庫(kù)的解決方案是以內(nèi)容驅(qū)動(dòng)分類基于技能的名稱、描述與正文提取關(guān)鍵詞通過確定性規(guī)則給出類別標(biāo)簽并把結(jié)果固化到skills_index.json中。當(dāng)前狀態(tài)來(lái)自 smart-auto-categorization.md倉(cāng)庫(kù)已通過生成的 catalog 完成索引大多數(shù)技能落入有意義類別仍有一小部分技能需要人工復(fù)審或補(bǔ)充關(guān)鍵詞覆蓋skills_index.json是類別標(biāo)簽與計(jì)數(shù)的唯一事實(shí)來(lái)源類別過濾器應(yīng)在構(gòu)建時(shí)從索引派生而非硬編碼文檔數(shù)字。從當(dāng)前索引實(shí)測(cè)以倉(cāng)庫(kù)內(nèi) skills_index.json 為準(zhǔn)共 2,124 條技能、106 個(gè)類別其中 development187、cloud146、ai-ml129、security86等居前仍有 333 條處于 uncategorized。這一數(shù)字會(huì)隨索引再生成而變化因此維護(hù)者文檔明確要求不要把固定計(jì)數(shù)復(fù)制進(jìn)用戶文檔一律從索引動(dòng)態(tài)生成。二、分類管線全景寫入即歸類整個(gè)自動(dòng)分類體系由兩條管線組成分工明確一次性回填工具auto_categorize_skills.py掃描倉(cāng)庫(kù)中尚無(wú)類別的技能直接改寫 SKILL.md 的 frontmatter索引生成器generate_index.py讀取所有 SKILL.mdfrontmatter 正文推斷類別并產(chǎn)出skills_index.json同時(shí)拷貝一份到 Web 應(yīng)用使用的apps/web-app/public/skills.json。另有配套測(cè)試 test_generate_index_categorization.py 驗(yàn)證推斷規(guī)則的正確性后面會(huì)詳細(xì)展開。三、分類規(guī)則詳解優(yōu)先級(jí)與打分機(jī)制3.1 優(yōu)先級(jí)系統(tǒng)分類決策遵循嚴(yán)格優(yōu)先級(jí)從源碼infer_category()見 generate_index.py可以確認(rèn)優(yōu)先級(jí)依據(jù)置信度reason 標(biāo)記1frontmatter 顯式category非 uncategorized1.0frontmatter:category2目錄結(jié)構(gòu)SKILL.md 所在父目錄0.95path:folder3關(guān)鍵詞匹配名稱描述正文0.45 0.05×score上限 0.92keyword-match:...4動(dòng)態(tài)推斷從技能 ID token 派生0.42 / 0.34 / 0.20derived-from-id-.../fallback:general原文檔簡(jiǎn)化為 Frontmatter category Detected Keywords Fallback (uncategorized)實(shí)際索引生成器將目錄結(jié)構(gòu)插在了第二級(jí)并且已不存在純 fallback 為 uncategorized 的情況——最后一級(jí)由動(dòng)態(tài)推斷兜底保證每個(gè)技能都拿到一個(gè)非空類別。3.2 基于關(guān)鍵詞的分析auto_categorize_skills.py中的categorize_skill()tools/scripts/auto_categorize_skills.py把技能名與描述拼接成小寫文本再為每個(gè)類別統(tǒng)計(jì)關(guān)鍵詞命中得分。3.3 作用域匹配與加權(quán)原文檔提到兩條關(guān)鍵規(guī)則源碼實(shí)現(xiàn)完全對(duì)應(yīng)精確短語(yǔ)匹配權(quán)重更高re.search(r\b re.escape(keyword) r\b, text)命中記 2 分auto_categorize_skills.py中而包含式匹配只記 1 分詞邊界避免誤報(bào)使用\b...\b正則防止 react 匹配到 reactive 之類的子串。索引生成器的generate_index.py將上述權(quán)重進(jìn)一步細(xì)化為詞邊界命中 3 分長(zhǎng)度 ≥5 的關(guān)鍵詞包含匹配 1 分且綜合id name description body_text四路文本見 skill_categorization/tools/scripts/generate_index.py。3.4 動(dòng)態(tài)推斷與停用詞當(dāng)關(guān)鍵詞全部未命中時(shí)infer_dynamic_category()會(huì)從技能 ID 拆 token若前兩個(gè) token 命中已知前綴集合azure、aws、google、github、gitlab、slack、discord、shopify、wordpress、odoo、notion、expo、react、nextjs、kubernetes 等生成${prefix}-${token}類別置信度 0.42否則取最后一個(gè) token 作類別置信度 0.34全部失敗則歸入general置信度 0.20。token 清洗會(huì)剔除STOPWORD_TOKENSskill、tools、builder、expert、guide、workflow、v2、ts、py 等 40 常見詞并丟棄長(zhǎng)度小于 3 的 tokenskill_categorization/tools/scripts/generate_index.py。四、動(dòng)手實(shí)踐運(yùn)行自動(dòng)分類4.1 針對(duì)未分類技能執(zhí)行分類python tools/scripts/auto_categorize_skills.py腳本會(huì)遍歷skills/目錄下每個(gè)含 SKILL.md 的目錄解析 frontmatterYAML 解析失敗則跳過并告警已有非 uncategorized 類別 → 計(jì)入already_categorized并跳過無(wú)類別 → 調(diào)用categorize_skill()打分最高分類別寫入 frontmatter無(wú)任何命中 → 計(jì)入failed_count。4.2 Dry-Run 預(yù)演推薦先跑python tools/scripts/auto_categorize_skills.py --dry-run--dry-run只計(jì)算并打印報(bào)告、不寫任何文件源碼中if not dry_run:才執(zhí)行 YAML 回寫見 auto_categorize_skills.py適合在正式回填前人工核對(duì)變更集。4.3 輸出報(bào)告解讀運(yùn)行結(jié)束后會(huì)打印如下結(jié)構(gòu)的報(bào)告 AUTO-CATEGORIZATION REPORT Summary: ? Categorized: 776 ?? Already categorized: 46 ? Failed to categorize: 124 Total processed: full repository Sample changes: ? 3d-web-experience uncategorized → web-development ? ab-test-setup uncategorized → testing ? agent-framework-azure-ai-py uncategorized → backend注意上述數(shù)字是文檔編寫時(shí)的樣例快照每次運(yùn)行結(jié)果取決于當(dāng)前倉(cāng)庫(kù)狀態(tài)原文檔也強(qiáng)調(diào)任何面向用戶的計(jì)數(shù)都應(yīng)以skills_index.json實(shí)況為準(zhǔn)。五、用索引實(shí)時(shí)查看類別分布不要手工統(tǒng)計(jì)或復(fù)制舊數(shù)字直接從索引生成這是原文檔給出的標(biāo)準(zhǔn)做法node - NODE const fs require(fs); const skills JSON.parse(fs.readFileSync(skills_index.json, utf8)); const counts new Map(); for (const skill of skills) { const category skill.category || uncategorized; counts.set(category, (counts.get(category) || 0) 1); } console.log(skills${skills.length} categories${counts.size}); for (const [category, count] of [...counts.entries()].sort((a, b) b[1] - a[1]).slice(0, 12)) { console.log(${category}: ${count}); } NODE在倉(cāng)庫(kù)根目錄執(zhí)行上述腳本可得到skills2124 categories106之類的實(shí)時(shí)輸出當(dāng)前實(shí)測(cè)數(shù)據(jù)隨倉(cāng)庫(kù)演進(jìn)會(huì)變化。分布信息還可通過generate_skills_report.py等配套工具進(jìn)一步沉淀見 tools/scripts。六、索引生成器類別如何落盤6.1 再生成索引python tools/scripts/generate_index.py執(zhí)行后依據(jù) tools/scripts/generate_index.py遍歷skills/跳過隱藏目錄僅處理含 SKILL.md 的目錄解析 frontmatter 并清洗含的未加引號(hào)值自動(dòng)補(bǔ)雙引號(hào)兼容郵箱等值frontmatter 字段name/description/risk/source/date_added/category優(yōu)先于目錄默認(rèn)值無(wú) description 時(shí)回退提取正文第一個(gè)非標(biāo)題段落截?cái)?250 字符調(diào)用infer_category()得到categorycategory_confidencecategory_reason按名稱排序后寫入skills_index.jsonUTF-8、LF 行尾并拷貝到apps/web-app/public/skills.json。6.2 索引中的分類字段實(shí)測(cè)skills_index.json中每條記錄包含id、path、category、name、description、risk、source、date_added、plugin等字段。其中category_confidence與category_reason是審計(jì)分類依據(jù)的利器——例如某技能 reason 為keyword-match:llm,embedding,rag即表示由這三個(gè)關(guān)鍵詞命中ai-ml。6.3 測(cè)試如何守護(hù)分類規(guī)則test_generate_index_categorization.py 覆蓋四條關(guān)鍵路徑normalize_category(Cloud DevOps)→cloud-devopskebab-case 歸一化frontmatter 顯式security→ 直接采用且置信度 1.0、reason 以frontmatter:開頭關(guān)鍵詞命中kubernetes/deployment→cloud-devops且置信度 ≥ 0.5無(wú)關(guān)鍵詞的vendorx-hyperflux→ 動(dòng)態(tài)推斷為hyperfluxderived-from-id-token:。這組測(cè)試保證了優(yōu)先級(jí)與置信度契約不會(huì)在后續(xù)重構(gòu)中悄悄退化。七、Web 應(yīng)用中的分類過濾分類的最終價(jià)值體現(xiàn)在apps/web-app的瀏覽體驗(yàn)源碼見 apps/web-app/src/pages/Home.tsx類別篩選器categoryFilter通過 URL query 參數(shù)?category驅(qū)動(dòng)支持分享與回退篩選邏輯為categoryFacet(skill.category) categoryFilter類別值直接取自索引下拉與按鈕列表從構(gòu)建期生成的索引動(dòng)態(tài)派生。改進(jìn)前后對(duì)比源自原文檔Before無(wú)序列表包含 uncategorized無(wú)類別規(guī)模提示After類別按技能數(shù)降序最多在前uncategorized 最后計(jì)數(shù)來(lái)自生成索引而非硬編碼文檔數(shù)字瀏覽體驗(yàn)顯著提升。下拉排序規(guī)則為All Categories → 最高計(jì)數(shù)類別 → 依次排列 → uncategorized若存在收尾。八、技能創(chuàng)作者指南讓新技能一次歸類8.1 新增技能時(shí)聲明類別在 SKILL.md 的 frontmatter 中直接寫入類別即可在下次索引生成時(shí)被優(yōu)先采用置信度 1.0--- name: my-skill description: ... category: web-development date_added: 2026-03-06 ---8.2 不確定類別怎么辦無(wú)需糾結(jié)下次重新生成索引時(shí)系統(tǒng)會(huì)自動(dòng)歸類。只需執(zhí)行python tools/scripts/generate_index.py生成器會(huì)依次走 frontmatter → 目錄結(jié)構(gòu) → 關(guān)鍵詞 → 動(dòng)態(tài)推斷的優(yōu)先級(jí)鏈絕不會(huì)把技能丟棄為未分類。九、內(nèi)置類別與關(guān)鍵詞速查表以下關(guān)鍵詞體系來(lái)自 auto_categorize_skills.py 的CATEGORY_KEYWORDS配置共 13 個(gè)類別覆蓋原文檔 Keyword Reference 全部條目并有所擴(kuò)展web-developmentreact, vue, angular, svelte, nextjs, gatsby, remix, html, css, javascript, typescript, frontend, web, tailwind, bootstrap, sass, less, webpack, vite, rollup, parcel, rest api, graphql, http, fetch, axios, cors, responsive, seo, accessibility, a11y, pwa, progressive, dom, jsx, tsx, component, router, routingbackendnodejs, node.js, express, fastapi, django, flask, spring, java, python, golang, rust, c#, csharp, dotnet, .net, laravel, php, ruby, rails, server, backend, api, rest, graphql, database, sql, mongodb, postgres, mysql, redis, cache, authentication, auth, jwt, oauth, session, middleware, routing, controller, modeldatabasedatabase, sql, postgres, postgresql, mysql, mariadb, mongodb, nosql, firestore, dynamodb, cassandra, elasticsearch, redis, memcached, graphql, prisma, orm, query, migration, schema, indexai-mlai, artificial intelligence, machine learning, ml, deep learning, neural, tensorflow, pytorch, scikit, nlp, computer vision, cv, llm, gpt, bert, classification, regression, clustering, transformer, embedding, vector, training, modeldevopsdevops, docker, kubernetes, k8s, ci/cd, git, github, gitlab, jenkins, gitlab-ci, github actions, aws, azure, gcp, terraform, ansible, vagrant, deploy, deployment, container, orchestration, monitoring, logging, prometheus, grafanacloudaws, amazon, azure, gcp, google cloud, cloud, ec2, s3, lambda, cloudformation, terraform, serverless, functions, storage, cdn, distributedsecuritysecurity, encryption, cryptography, ssl, tls, hashing, bcrypt, jwt, oauth, authentication, authorization, firewall, penetration, audit, vulnerability, privacy, gdpr, compliancetestingtest, testing, jest, mocha, jasmine, pytest, unittest, cypress, selenium, puppeteer, e2e, unit test, integration, coverage, ci/cdmobilemobile, android, ios, react native, flutter, swift, kotlin, objective-c, app, native, cross-platform, expo, cordova, xamaringame-developmentgame, unity, unreal, godot, canvas, webgl, threejs, babylon, phaser, sprite, physics, collision, 2d, 3d, shader, renderingdata-sciencedata, analytics, science, pandas, numpy, scipy, jupyter, notebook, visualization, matplotlib, plotly, statistics, correlation, regression, clusteringautomationautomation, scripting, selenium, puppeteer, robot, workflow, scheduled, trigger, integrationcontentmarkdown, documentation, content, blog, writing, seo, meta, schema, og, twitter, description需要說(shuō)明索引生成器 skill_categorization/tools/scripts/generate_index.py 維護(hù)的是另一套精簡(jiǎn)類別名web-engineering、cloud-devops、testing-qa、data-engineering、research、bioinformatics、geospatial、finance 等兩者用途不同——前者面向一次性的 SKILL.md 回填后者面向持續(xù)性的索引推斷。修改任何一方后都應(yīng)重新跑對(duì)應(yīng)流程以保持同步。十、自定義擴(kuò)展新增類別與關(guān)鍵詞10.1 添加自定義關(guān)鍵詞編輯 tools/scripts/auto_categorize_skills.py 的CATEGORY_KEYWORDS配置CATEGORY_KEYWORDS { your-category: [ keyword1, keyword2, exact phrase, another-keyword ], # ... other categories }然后按順序重新執(zhí)行兩條命令讓回填與索引保持一致python tools/scripts/auto_categorize_skills.py python tools/scripts/generate_index.py10.2 類別歸一化規(guī)則所有類別值統(tǒng)一歸一化為小寫 kebab-casenormalize_category()下劃線轉(zhuǎn)連字符、空白折疊、非法字符剔除、連續(xù)連字符合并。因此寫Cloud DevOps、cloud_devops都會(huì)收斂為cloud-devops避免同類別多標(biāo)簽分裂。十一、故障排查與最佳實(shí)踐11.1 仍有 Failed to categorize 的技能原文檔給出的三種處置路徑按推薦順序在 frontmatter 手動(dòng)指定類別category: your-chosen-category向CATEGORY_KEYWORDS補(bǔ)充關(guān)鍵詞讓下一次自動(dòng)分類能命中移動(dòng)到更貼合的目錄目錄結(jié)構(gòu)會(huì)作為第二優(yōu)先級(jí)回退skills/backend/my-new-skill/SKILL.md此外建議檢查是否命中了停用詞或描述過短——索引生成器對(duì)缺失 description 的技能會(huì)回退抽取正文首段因此為每個(gè)技能寫一段信息豐富的 description 是提升分類質(zhì)量最經(jīng)濟(jì)的手段。11.2 修改 SKILL.md 后重新生成索引python tools/scripts/generate_index.py該命令將依次完成解析 frontmatter 類別 → 回退目錄結(jié)構(gòu) → 關(guān)鍵詞/動(dòng)態(tài)推斷 → 生成新的skills_index.json→ 拷貝至apps/web-app/public/skills.json。十二、后續(xù)演進(jìn)路線原文檔列出的后續(xù)工作可歸納為四件事在 Web 應(yīng)用中驗(yàn)證改進(jìn)后的類別過濾器排序與計(jì)數(shù)是否與索引一致補(bǔ)齊缺失關(guān)鍵詞對(duì)仍 uncategorized 的技能逐一補(bǔ)充命中詞人工復(fù)核剩余未分類技能自動(dòng)指派或人工審查雙軌并行持續(xù)監(jiān)控增長(zhǎng)利用分類報(bào)告跟蹤新增 vs 已分類的變化趨勢(shì)。延伸閱讀分類設(shè)計(jì)完整方案docs/SMART_AUTO_CATEGORIZATION.md、docs/CATEGORIZATION_IMPLEMENTATION.md索引與目錄結(jié)構(gòu)說(shuō)明docs/README.md、CATALOG.md中文對(duì)照版維護(hù)指南docs_zh-CN/maintainers/smart-auto-categorization.md核心實(shí)現(xiàn)tools/scripts/auto_categorize_skills.py、tools/scripts/generate_index.py、skill_categorization/tools/scripts/generate_index.py規(guī)則測(cè)試skill_categorization/tools/scripts/tests/test_generate_index_categorization.pyWeb 應(yīng)用篩選實(shí)現(xiàn)apps/web-app/src/pages/Home.tsx【免費(fèi)下載鏈接】agentic-awesome-skillsAAS Core is the local, agent-first control plane for complete catalog discovery, agent-owned selection, stack validation, and planning, backed by 2,115 agentic skills. Includes CLI, local MCP, catalog, plugins, and Workbench.項(xiàng)目地址: https://gitcode.com/gh_mirrors/an/agentic-awesome-skills創(chuàng)作聲明：本文部分內(nèi)容由AI輔助生成（AIGC），僅供參考