Ukwakha uhlelo lokuphatha oluqhutshwa yi-ontology kusukela ezidingweni zebhizinisi usebenzisa i-ontology-driven-dev

Isingeniso Sephrojekthi

ontology-driven-dev iyikhono le-AI eliqondiswe ekuthuthukiseni amasistimu ebhizinisi, lisebenzisa indlela eqinile enezinyathelo ezintathu ngokulandelana: “Ukuhlola izidingo → Ukwenza imodeli ye-ontology → Ukwakha uhlelo lokusebenza”. Ungaqala ngencazelo eyodwa yebhizinisi, bese kuthi ngemva kokuqinisekiswa komuntu esigabeni ngasinye uthole incwadi yezincazelo zezidingo, ama-YAML amamodeli ayisikhombisa, kanye nesistimu esebenzayo yewebhu enezingxenye zangaphambili nezingemuva.

Inhloso yayo eyinhloko akukhona nje ukwenza i-AI ikhiqize ikhodi ngokushesha, kodwa ukuqinisekisa ukuthi imibhalo yezidingo, imodeli ye-ontology kanye nekhodi yokugcina kuhambisana ngokuqinile. Amathebula esizindalwazi, izindawo zokuxhumana, amamenyu, izimvume, izinqubo, imithetho nam报告 konke kusebenzisa imodeli njengomthombo wencazelo, ngaleyo ndlela kuncishiswe inkinga yokungahambisani kwezidingo, umklamo nokusetshenziswa.

Isigaba sokuhlola izidingo sinezinyathelo eziyisishiyagalombili. Isinyathelo ngasinye kufanele simiswe kulindwe ukuqinisekiswa komuntu, futhi i-Agent ayikwazi ukusidlula ngokuzenzakalelayo.

Amakhono Abalulekile Nokwakheka Kwephrojekthi

Izici Ezisemqoka

• Ukulandeleka kwezidingo: Umsebenzi ngamunye ungabuyiselwa entweni ehambisanayo encwadini yezincazelo zezidingo.
• Incazelo ehlanganisiwe yamamodeli ayisikhombisa: I-M1, M2, M3, M5, M6, M7 kanye ne-MU YAML kusetshenziswa ukuchaza izinto, ukuziphatha, imithetho, abathintekayo, izinqubo, imibuzo nemibiko kanye ne-UI.
• Isango lokuqinisekisa komuntu: Izigaba zero kuya kweziyisikhombisa zokuhlola izidingo zidinga ukuqinisekiswa komuntu ngaphambi kokuqhubeka.
• Isisekelo sobuchwepheshe esakhelwe ngaphakathi: code-paas inikeza isisekelo sohlelo lokusebenza olulodwa olusebenzisa i-Flask, i-SQLite, i-React ne-TypeScript, futhi iqukethe ukuphathwa kwesistimu, injini yezinqubo, ideshibhodi yokusebenza kanye nerejista ye-ontology.
• Ukusebenzisana okuzalwa yi-AI: Uhlelo olwakhiwe kufanele lube nebhokisi lengxoxo le-AI ngakwesokudla, elisekela ukufakwa kwerejista ye-ontology, ukubizwa kwamathuluzi, impendulo esakazwa nge-SSE kanye nomngcele wokuphepha we-SQL efundwa kuphela.
• Ukuhambisana namathuluzi amaningi: Ingasetshenziswa ku-WorkBuddy, Claude Code, Codex, Cursor, Aider, Cline kanye nanoma iyiphi i-Agent evamile ekwazi ukulayisha imiyalelo yephrojekthi.

Okuqukethwe Okuyinhloko Kurepositri

• SKILL.md: Ichaza indlela yokusebenza, iphayiphi enezigaba ezintathu kanye nemithetho yokwenza okufanele ilandelwe.
• references/: Iqukethe amazinga okubhala izidingo, uhlaka lokwenza imodeli ye-ontology, umhlahlandlela wokuthuthukisa osuka kumodeli uye ekusetshenzisweni, ukwakheka kobuchwepheshe bokusebenzisana okuzalwa yi-AI kanye namazinga e-UI/UE.
• reference-example/: Isibonelo esiphelele sokuphathwa kokwenziwa kwezinkontileka zokuthengisa, esihlanganisa imibhalo yezidingo, ama-YAML amamodeli ayisikhombisa kanye ne-manifest.json.
• techbase/: Ikhodi yomthombo ye-code-paas engakopishwa futhi inwetshwe, ehlanganisa izingxenye ezingemuva, zangaphambili, izibonelo zamamodeli namafayela ezincike kuwo.

Ukulungisa Indawo Yokusebenza

Ngaphambi kokuqala, qinisekisa ukuthi ikhompuyutha yakho inale ndawo elandelayo:

• Python 3.10 noma inguqulo entsha.
• Node.js 18 noma inguqulo entsha.
• Ukufinyelela kunethiwekhi lapho kufakwa okokuqala ukuncika kwe-Python nokwangaphambili.

Ingxenye engemuva isebenzisa i-Flask ne-SQLite, kanti ingxenye yangaphambili isebenzisa i-React, i-TypeScript ne-Vite. Iphrojekthi isebenzisa i-MIT License, ngakho ingasetshenziswa, iguqulwe futhi isatshalaliswe kabusha ngokukhululeka.

Ukufaka Kumathuluzi Okuthuthukisa e-AI

Ukufaka ku-Claude Code

I-Claude Code ingathola ngokuzenzakalelayo i-SKILL.md enkombeni yamakhono. Ungakhetha ukufaka ezingeni lomsebenzisi ukuze wonke amaphrojekthi akwazi ukuyisebenzisa:

cp -r ontology-driven-dev ~/.claude/skills/ontology-driven-dev

Ungayifaka futhi kuphrojekthi yamanje:

cp -r ontology-driven-dev .claude/skills/ontology-driven-dev

Ngemva kokuqeda, i-Claude Code izobona futhi ilayishe inqubo ngokusekelwe ku-name kanye ne-description okukwi-frontmatter yekhono.

Ukufaka ku-WorkBuddy

Umyalo wokufaka ezingeni lomsebenzisi ulandelayo:

cp -r ontology-driven-dev ~/.workbuddy/skills/ontology-driven-dev

Uma ufuna ukuyisebenzisa kuphrojekthi eyodwa kuphela, sebenzisa:

cp -r ontology-driven-dev <iphrojekthi_yakho>/.workbuddy/skills/ontology-driven-dev

Ukufaka ku-Codex

I-Codex ayinayo irejista yamakhono eyakhelwe ngaphakathi, ngakho qala ngokukopisha ikhono uye kurepositri:

mkdir -p .codex/skills
cp -r ontology-driven-dev .codex/skills/

Bese ufaka lo mthetho wokulayisha ku-codex.md noma ku-AGENTS.md yephrojekthi:

Lapho umsebenzisi efuna ukwenza “ukuthuthukiswa okuqhutshwa yi-ontology, ukuhlola izidingo, ukwenza imodeli ye-ontology, amamodeli ayisikhombisa, i-code-paas noma uhlelo lokusebenza oluzalwa yi-AI”, layisha .codex/skills/ontology-driven-dev/SKILL.md, bese ulandela ngokuqinile inqubo enezigaba ezintathu kanye nesango lokuqinisekisa komuntu.

I-Codex ihumusha ukuqinisekiswa komuntu kube imibuzo esebenzisanayo noma ukugunyazwa. Isandbox yayo idinga uxhumano lwenethiwekhi ukuze ikwazi ukusebenzisa npm install kanye ne-pip install.

Ukusetshenziswa ku-Cursor, Aider noma Cline

• Cursor: Faka okuqukethwe kwe-SKILL.md emithethweni yephrojekthi noma ku-.cursorrules.
• Cline, Aider: Namathisela i-SKILL.md ekuqaleni kweseshini, noma ucele ngokusobala i-Agent ukuthi ilayishe indlela yayo.
• Amanye ama-Agent: Faka leli fayela kumongo njengomyalelo wesistimu noma inkumbulo yephrojekthi.

Isigaba Sokuqala: Ukuhlola Nokuqinisekisa Izidingo

Isigaba sokuqala sisekelwe ku-references/AI需求探索与确认提示词V9.0.md, futhi sihlela incazelo yokuqala yebhizinisi ibe incwadi esemthethweni yezincazelo zezidingo zesofthiwe.

Ungafaka umyalo ofana nalona ku-Agent:

Ngisize ngakhe uhlelo lokuphathwa kokwenziwa kwezinkontileka zokuthengisa, oludinga ukuphatha izinkontileka, izinkokhelo ezitholiwe, ukulethwa nokugunyazwa.

I-Agent kufanele iqhubekele phambili ngokulandelana kwalezi zigaba eziyisishiyagalombili:

1. Ukuqonda jikelele.
2. Izinto zebhizinisi.
3. Imisebenzi nemithetho yebhizinisi.
4. Ukuxhumana phakathi kwezinto.
5. Ukusebenzisana nokugeleza kokugunyazwa kusukela ekuqaleni kuya ekugcineni.
6. Imibuzo, izibalo nemibiko.
7. Izindima nezimvume.
8. I-prototype ye-UI, engakhethwa ngokwesidingo.

Ngemva kokuphela kwesigaba ngasinye, i-Agent kufanele iqale ukuqinisekisa ngefomethi ethi “Umbuzo, isiphakamiso se-AI, isizathu, ezinye izinketho, impendulo esheshayo”, bese ima ilinde isinqumo sakho. Okuqukethwe kohlobo B okuqondene nebhizinisi nakho kufanele kubuzwe ngokucacile; ngaphandle kokuqinisekiswa, akuvunyelwe ukuya esigabeni esilandelayo.

Ekugcineni kwalesi sigaba kukhiqizwa:

<isizinda_sebhizinisi>-需求规格说明书-V9.md

Isithasiselo C salo mbhalo siyisisekelo sokufaka amamodeli ayisikhombisa esigabeni esilandelayo, ngakho kufanele kuhlolwe ngokukhethekile ukuthi izinto zebhizinisi, izinqubo, imithetho, izindima nemibiko ziphelele yini.

Isigaba Sesibili: Ukukhiqiza ama-YAML Amamodeli Ayisikhombisa

Ngemva kokuqinisekiswa kwezidingo, i-Agent yakha imodeli ye-ontology ngokusekelwe ku-references/ontology_modeling_framework_v9.md kanye nasesithasiselweni C sombhalo wezidingo. Ngalesi sikhathi okokufaka sekuyisisekelo esiqinisekisiwe, ngakho akufanele kuqale kabusha ukuhlukanisa ibhizinisi ngezinga elikhulu.

Uma usuvele unencwadi yezincazelo zezidingo ehambisana nezimfuneko, ungavula ngokuqondile inqubo yokwenza imodeli kuphela:

Ngokusekelwe kule ncwadi yezincazelo zezidingo, yenza imodeli ye-ontology bese ukhipha ama-YAML amamodeli ayisikhombisa kanye ne-manifest.json.

Okuqukethwe okukhiqizwa enkombeni ye-yaml/ kuhlanganisa:

• Imodeli yezinto i-M1: Ichaza izinto zebhizinisi.
• Imodeli yokuziphatha i-M2: Ichaza ukuziphatha kwebhizinisi.
• Imodeli yemithetho i-M3: Ichaza imithetho yebhizinisi.
• Imodeli yabathintekayo i-M5: Ichaza izindima nabathintekayo abahlobene.
• Imodeli yenqubo i-M6: Ichaza izinqubo zebhizinisi.
• Imodeli yemibuzo nemibiko i-M7: Ichaza imibuzo nemibiko.
• Imodeli ye-UI i-MU: Ichaza ukuxhumana komsebenzisi.
• manifest.json: Igcina uhlu lwamamodeli.

Ngemva kokuqeda ukwenza imodeli, kufanele kwenziwe ukuhlolwa kokuvumelana, okungenani kubandakanye ukulandeleka kwezidingo, ubudlelwano obuqondile phakathi kwe-M7 ne-M2, kanye nokuqinisekisa ukuthi izinkomba ze-M6 azinawo umjikelezo. Uma kutholakala ukungahambisani, lungisa imodeli kuqala ngaphambi kokungena esigabeni sokwakha ikhodi.

Isigaba Sesithathu: Ukwakha Uhlelo Lwe-BS Olusebenzayo

Ukukopisha Isisekelo Sobuchwepheshe se-code-paas

Kopisha ngokuphelele i-techbase/ ekhoneni lamakhono uye ku-code-app/ yephrojekthi yamanje:

cp -r <impande_yekhono>/techbase/. <iphrojekthi_yamanje>/code-app/
cd <iphrojekthi_yamanje>/code-app/frontend && npm install
cd <iphrojekthi_yamanje>/code-app/backend && pip install -r requirements.txt

Lapho uthuthukisa ngokusemthethweni, kufanele kusetshenziswe ama-YAML amamodeli ayisikhombisa akhiqizwe esigabeni sesibili esikhundleni sezibonelo ezisesisekelweni.

Ukusetshenziswa Ngokulandelana Kwezinyathelo Eziyishumi

1. Bhala noma uhlele ama-YAML amamodeli ayisikhombisa.
2. Khiqiza i-DDL namathebula esizindalwazi.
3. Bhalisa isichazamazwi sedatha.
4. Qalisa izinsizakalo zokuziphatha nemithetho.
5. Bhala idatha yokuqala yezindima nezimvume.
6. Xhuma injini yezinqubo.
7. Khiqiza amamenyu, amakhasi nemizila.
8. Qalisa ibhokisi lengxoxo le-AI elingakwesokudla.
9. Yenza ukuhlolwa okuphelele kokuxhumana phakathi kwezingxenye zangaphambili nezingemuva.
10. Yenza ukwamukelwa ngokusekelwe ezidingweni nasemamodelini.

Ibhokisi lengxoxo le-AI elingakwesokudla liyingxenye eyimpoqo futhi akufanele lishiywe ngesikhathi sokuthuthukisa. Kufanele lilandele ukufakwa kwerejista ye-ontology, ukubizwa kwamathuluzi, ukudluliswa kwe-SSE kanye nomngcele wokuphepha we-SQL efundwa kuphela ochazwe yiphrojekthi.

Ukuqalisa Izingxenye Ezingemuva Nezingaphambili

Qalisa ingxenye engemuva ye-Flask:

cd code-app/backend
pip install -r requirements.txt
python app.py

Ikheli elizenzakalelayo lokufinyelela engxenyeni engemuva ngu-http://localhost:5000.

Vula enye itheminali ukuze uqalise iseva yokuthuthukisa engaphambili:

cd code-app/frontend
npm install
npm run dev

Ikheli elizenzakalelayo lokufinyelela engxenyeni yangaphambili ngu-http://localhost:5173. Igama lomsebenzisi elizenzakalelayo ngu-admin, kanti iphasiwedi ezenzakalelayo ngu-admin123; eminye imininingwane yokusebenzisa isisekelo ingatholakala ku-techbase/README.md.

Izindlela Ezivamile Zokuqalisa

Ngokusekelwe kokokufaka okukhona nenhloso, ungasebenzisa amagama ahlukene okuqalisa.

Ukuqala Ngezidingo Nokuthuthukisa Ngokuphelele

Ngisize ngakhe uhlelo lokuphathwa kokulungiswa kwemishini, ngilandele ngokuqinile inqubo yokuthuthukisa eqhutshwa yi-ontology, futhi ulinde ukuqinisekiswa kwami esigabeni ngasinye sokuhlola izidingo.

Ukwenza Imodeli Ye-Ontology Kuphela

Yenza imodeli ye-ontology ngokusekelwe kule ncwadi yezincazelo zezidingo, bese ukhipha i-M1, M2, M3, M5, M6, M7, MU kanye ne-manifest.json.

Ukwakha Uhlelo Ngokusekelwe Kumamodeli Akhona

Khiqiza uhlelo ngokusekelwe kulawa ma-YAML amamodeli ayisikhombisa, sebenzisa isisekelo se-code-paas, bese uqedela ukuhlolwa kokuxhumana phakathi kwezingxenye zangaphambili nezingemuva.

Amagama angukhiye athi “ukuthuthukiswa okuqhutshwa yi-ontology”, “ukuhlola izidingo”, “ukwenza imodeli ye-ontology”, “amamodeli ayisikhombisa”, “code-paas”, “uhlelo lokusebenza oluzalwa yi-AI” kanye “nokuthuthukiswa kwezinhlelo zebhizinisi” angasiza futhi i-Agent ukuthi ibone ikhono.

Izincomo Zokusebenzisa Ezisezingeni Elithuthukile

Ungalidluli Isango Lokuqinisekisa Komuntu

Ngisho noma izidingo zokuqala zibonakala ziphelele, ungavumeli i-Agent ukuthi ikhiqize uhlelo lonke ngesikhathi esisodwa. Ukuqinisekisa isigaba nesigaba kusiza ukuveza kusenesikhathi izinkinga zemingcele yezinto, izindawo zokugunyaza, ububanzi bezimvume kanye nendlela yokubala izibalo.

Sebenzisa Isithasiselo C Njengomngcele Wokwenza Imodeli

Isigaba sesibili kufanele sisebenzise Isithasiselo C sombhalo wezidingo njengokokufaka okuqinisekisiwe. Uma kutholakala ukushoda kokuqukethwe kwebhizinisi okubalulekile ngesikhathi sokwenza imodeli, buyela esigabeni sezidingo ukuze wengeze futhi uqinisekise, kunokuba imodeli izandisele yona ububanzi bebhizinisi obungakaqinisekiswa.

Bheka Kuqala Izinkomba Eziphakathi Kwamamodeli

Ngaphambi kokuqala ukwakha ikhodi, qinisekisa ngokukhethekile ubudlelwano phakathi kwe-M7 ne-M2, ubudlelwano bezinkomba be-M6, kanye nokulandeleka kwezidingo kuya ezintweni zemodeli. Amaphutha asezingeni lemodeli angangena kusizindalwazi, kuma-interface nasemakhasini, okwandise izindleko zokuhlanganisa nokuhlola kamuva.

Gcina isibonelo esiyisisekelo ukuze usisebenzise ekuqhathaniseni

reference-example/ sinikeza isibonelo esiphelele sangempela sokuphathwa kokwenziwa kwezinkontileka zokuthengisa. Lapho usisebenzisa okokuqala, ungaqhathanisa naso incwadi yezincazelo zezidingo, ama-YAML amamodeli ayisikhombisa kanye ne-manifest.json, ukuze uqonde indlela imikhiqizo yezigaba ezihlukahlukene exhumana ngayo.

Shintsha amamodeli kuqala, bese wandisa isisekelo

Okuku-techbase/models/ ngamamodeli ayizibonelo. Kumaphrojekthi asemthethweni, qala ngokufaka ama-YAML akho amamodeli ayisikhombisa, bese wandisa izinsizakalo zangemuva, izimvume, izinqubo, amamenyu namakhasi ngokulandela uchungechunge lwezinyathelo eziyishumi olusuka kumodeli luya ekusetshenzisweni. Gwema ukulungisa ikhodi yebhizinisi ngokuqondile uzungeze idatha yesibonelo.

Isiphetho

i-ontology-driven-dev ixhumanisa ukucaciswa kwezidingo, ukumodelwa kwe-ontology kanye nokusetshenziswa kohlelo kube uchungechunge lokuthuthukisa olunamasango okuqinisekiswa abantu. Okubalulekile ekuyisebenziseni kahle ukuphikelela ekuqinisekiseni izidingo kuqala, bese uhlola ukuvumelana kwamamodeli ayisikhombisa, ekugcineni wakhe futhi wamukele uhlelo usebenzisa i-code-paas. Ngale ndlela ungasebenzisa i-AI ukuthuthukisa ukusebenza kahle kokuthuthukiswa kwezinhlelo zebhizinisi, kuyilapho ugcina umthombo ocacile wencazelo kanye nobudlelwano obuphelele bokulandelela izidingo.