Isifundo Somqoqi Wamakhadi E-McDonald’s: Sebenzisa i-WorkBuddy Ukudweba Amakhadi, Ukuhlola Umsoco Nokufaka I-oda Ngokuphephile

Uhlolojikelele lwephrojekthi

Umqoqi Wamakhadi e-McD (mcd-gacha-collector) yi-WorkBuddy Skill eyakhiwe kusetshenziswa i-MCP kaMcDonald's China. Iguqula ukudla kwaseMcDonald's kube amakhadi angadonswa futhi aqoqwe, libonise intengo, amakhalori, amaprotheni, i-sodium nezinga lezempilo ekhadini.

Le phrojekthi ithola imenyu yesikhathi sangempela, idatha yokudla, izitolo namanani nge-MCP, bese injini yendawo ye-Python yenza ukuhlola kwezempilo, ukuhlukanisa ngokungavamile, ukuhlunga ngesabelomali kanye nokudonsa amakhadi okunesisindo. Ngemva kokudonsa ukudla okusekela ukuthengwa, ingadala i-oda langempela ngemva kokubheka intengo nokuqinisekisa okwesibili.

Idatha yokudla ekwindawo yokugcina ihlanganisa cishe izinto zokudla ezingu-160, kanti ichibi lamakhadi lewebhu njengamanje linamakhadi angu-53 akhiqizwe ngolwazi lwemenyu yangempela. Le phrojekthi ingumsebenzi wokuncintisana, akuwona umkhiqizo osemthethweni kaMcDonald's.

Imisebenzi eyinhloko

• Ukudonsa ikhadi ngesabelomali: Ungenza izicelo zolimi lwemvelo ezifana nokuthi “donsa ikhadi elilodwa ngaphansi kwama-yuan angu-50”; uhlelo luqala luhlunge ngentengo, bese ludonsa ngokunesisindo.
• Ukudla nezinga lezempilo: Ngokusekelwe kudatha ebuyiswe yi-MCP, kubalwa izinga lezempilo elisuka ku-0 kuye ku-100, kuboniswe amakhalori, amaprotheni, amafutha noma i-sodium.
• Ukuhlukaniswa kwe-SSR, UR ne-R: Ukungavamile kunqunywa ukusatshalaliswa kwezinga lezempilo, akucaciswa ngesandla. Ukuhlukaniswa okuhlobene kuvumelana nesakhiwo semenyu sezitolo ezahlukene.
• I-albhamu yendawo yamaqoqo: Amarekhodi okudonsa agcinwa ku-ledger.json; amakhadi aphindaphindiwe aqongelela inani lokutholwa, futhi kungabalwa amakhalori amaphakathi nezinga lezempilo elimaphakathi.
• Indlela yamaphuzu: I-Skill ibuyisa ngokuzenzakalelayo ithuba lokudonsa eli-1 njalo emahoreni angu-4; inguqulo yewebhu isebenzisa ukuvuselela ngesandla okunokulinda kwemizuzwana engu-60.
• Uku-oda kwangempela: Iqala ngokubiza i-MCP ukuze ihlole intengo, bese icela ukuqinisekiswa okucacile komsebenzisi, futhi ekugcineni ibize i-create-order.
• Ukusetshenziswa ngaphandle kwe-inthanethi: Indawo yokugcina ihlanganisa idatha yesampula, ngakho ungahlola ukudonsa amakhadi, amaqoqo nokunikezwa kwamakhadi e-HTML ngaphandle kokulungisa i-MCP Token.
• Akukho ukuncika komuntu wesithathu: Izikripthi eziyinhloko zisebenzisa umtapo ojwayelekile we-Python, futhi zidinga i-Python 3.8 noma ngaphezulu.

Indlela yokusebenza

Le phrojekthi ilandela umqondo wokuthi “idatha inikezwe i-MCP, izibalo zenziwe yikhodi, ukuvezwa kwenziwe yi-Agent”. I-WorkBuddy Agent iqonda inhloso yomsebenzisi, i-McDonald's MCP inikeza idatha yebhizinisi yangempela, bese i-gacha.py yenza izibalo ezinqunyiwe.

Umsebenzisi wenza isicelo sokudonsa ikhadi, ukuhlola ukudla noma uku-oda
→ I-WorkBuddy Agent ibiza i-McDonald's MCP
→ Ithola isikhathi, isitolo, imenyu, intengo nedatha yokudla
→ i-gacha.py ibala izinga lezempilo nokungavamile, bese idonsa ikhadi
→ i-render.py ikhiqiza ikhadi noma ikhasi le-albhamu yamaqoqo
→ Umsebenzisi ukhetha ukuthenga
→ I-calculate-price ihlola intengo
→ Umsebenzisi uqinisekisa okwesibili
→ I-create-order idala i-oda

Imali, izinga lezempilo nokungavamile akubalwa ngomlomo yimodeli enkulu yolimi. Intengo ibuya ku-MCP, kanti ukuhlola nokudonsa amakhadi kwenziwa yikhodi ye-Python, ngaleyo ndlela kuncishiswe ubungozi bokuphambuka nokwenza amaphutha ekubaleni.

Okulungiselelwa ngaphambi kokuqala

• Faka i-Python 3.8 noma ngaphezulu.
• Faka futhi ukwazi ukusebenzisa i-WorkBuddy.
• Lungisa i-McDonald's China MCP Token efinyeleleka; i-Token ayidingeki uma usebenzisa izibonelo ezingaxhunyiwe ku-inthanethi kuphela.
• Qinisekisa ukuthi i-WorkBuddy ingafunda ifolda yamakhono yomsebenzisi.

Indlela yokuqala: zama ngqo ku-inthanethi

Uma ufuna ukubuka isixhumi esibonakalayo nomboniso wokudonsa amakhadi kuphela, vula ikhasi le-inthanethi loMqoqi Wamakhadi e-McD. Ikhasi lizozama ukuthola idolobha nge-IP, bese libiza umsebenzi wokusesha izitolo ukuze likhethe isitolo esiseduze, ngaphandle kokufaka ikheli ngesandla.

Ngenxa yemikhawulo yendawo yokuthunyelwa, ikhasi lewebhu lomphakathi alikwazi ukufaka i-MCP Token yomuntu siqu. Inguqulo ye-inthanethi ilungele ukuzama indawo, izitolo nesixhumi esibonakalayo samakhadi; uku-oda kwangempela kufanele kwenziwe kuseva yendawo noma ku-WorkBuddy Skill enesixhumi esilungisiwe.

Indlela yesibili: faka ku-WorkBuddy

Isinyathelo 1: klona indawo yokugcina

git clone https://github.com/YOUR_GITHUB_USER/mcd-gacha-collector.git
cd mcd-gacha-collector

Ikheli lendawo yokugcina ku-README lisebenzisa isibambi-ndawo. Ekusebenziseni kwangempela, shintsha i-YOUR_GITHUB_USER ufake igama lomnikazi wendawo yokugcina, noma ukopishe ngqo ikheli lokuklona elinikezwa yikhasi le-GitHub.

Isinyathelo 2: cela i-McDonald's MCP Token

Vakashela inkundla evulekile ye-McDonald's MCP, ngena ngenombolo yeselula, uvule isevisi kukhonsoli bese ukopisha i-Token. I-Token iyimininingwane ebucayi; ungayibhali endaweni yokugcina yomphakathi futhi ungayifaki emlandweni we-Git.

Isinyathelo 3: lungisa isixhumi se-MCP

1. Vula “专家·技能·连接器” ngakwesokunxele ku-WorkBuddy.
2. Ngena ku-“连接器”, bese ukhetha “自定义连接器” phezulu kwesokudla.
3. Khetha “配置 MCP”.
4. Kopisha okuqukethwe kwe-mcp-config.example.json endaweni yokugcina.
5. Shintsha i-${MCD_MCP_TOKEN} ufake i-Token yangempela, noma uyifake ngokuguquguquka kwemvelo.
6. Londoloza ukulungiselela bese uvula isixhumi.

Kuyanconywa ukusebenzisa ukuguquguquka kwemvelo ukuphatha i-Token. Noma iyiphi indlela oyisebenzisayo, akufanele ufake ifayela lokulungiselela eliqukethe imininingwane yangempela ohlelweni lokulawula izinguqulo.

Isinyathelo 4: kopisha i-Skill

Ku-macOS noma ku-Linux sebenzisa:

cp -r skills/mcd-gacha-collector ~/.workbuddy/skills/

Ifolda yamakhono yomsebenzisi ye-WorkBuddy ku-Windows yile:

%USERPROFILE%\.workbuddy\skills\

Ngesikhathi sokukopisha, gcina yonke ifolda ye-mcd-gacha-collector, ehlanganisa i-SKILL.md, izikripti eziyinhloko, imibhalo eyisethenjwa namafayela ezinsiza.

Isinyathelo 5: qinisekisa ukufakwa

Buyela ebhokisini lengxoxo le-WorkBuddy, bese ufaka:

抽张麦当劳卡

Uma i-Skill ingabiza amathuluzi emenyu nawokudla futhi ibuyise ikhadi elinokungavamile, intengo namakhalori, ukufakwa nokuxhumeka kwe-MCP sekusebenzile.

Indlela eyisisekelo yokusebenzisa

Donsa ikhadi elilodwa ngesabelomali

50 元内抽一张

I-Skill ithola imenyu etholakalayo nedatha yokudla yesitolo samanje, ihlunge ukudla okunentengo engeqi ama-yuan angu-50, bese idonsa ikhadi ngokunesisindo. Isibonelo esihloliwe ku-README sidonse UR 阳光柠檬红茶中杯, ngentengo engu-10 yuan, amakhalori angu-117 kcal nezinga lezempilo elingu-63.4.

Umphumela wangempela uncike kumenyu yesitolo, amanani esikhathi sangempela, isakhiwo sechibi lamakhadi kanye nomphumela wokudonsa okungahleliwe, ngakho ukudla noma intengo esibonelweni akufanele kubhekwe njengomphumela ongaguquki.

Yenza ukudonsa okuyishumi

来十连

Uma kunamaphuzu anele, i-Skill izodonsa amakhadi angu-10 ngokulandelana futhi ibonise imiphumela ngendlela ephelele. Uma amaphuzu enganele, izochaza ukuthi mangaki amathuba asashodayo nokuthi ithuba elilandelayo lizobuya nini.

Buka i-albhamu yamaqoqo

我的收藏册？

Idatha yamaqoqo ihlanganiswa yi-ledger.json yendawo. Imiphumela ingafaka inani lamakhadi aqoqiwe, amanani e-SSR, UR ne-R, inani lokuphindaphinda, amakhalori amaphakathi nezinga lezempilo elimaphakathi.

Buza ngolwazi lokudla

巨无霸多少热量？

I-Skill isebenzisa i-list-nutrition-foods ukuthola ukudla. Umphumela oyisibonelo ku-README uthi: i-Big Mac inamakhalori angu-513, amaprotheni angu-27 g, amafutha angu-26 g, i-sodium engu-961 mg, nezinga lezempilo elingu-32.2, futhi ihlukaniswe ku-R. Ulwazi lokudla kufanele lusasekelwe kudatha esemthethweni ebuyiswe yi-MCP ngaleso sikhathi.

Thola izitolo eziseduze

附近麦当劳？

Uhlelo luqala ngokuthola indawo nge-IP, luthole idolobha, bese lubiza i-query-nearby-stores futhi luhlele ngokwenkambu yebanga ebuyiswe yi-MCP. Njengoba izixhumanisi zesitolo kungenzeka zihlale zingu-0.0 ekuhlolweni kwangempela, iphrojekthi ayincikile kulezi zixhumanisi ukubala ibanga ngokwayo.

Buka amaphuzu asele

还有几次抽卡机会？

I-Skill ifunda amaphuzu ku-ledger yendawo, bese ibuyisa inani lamanje lamathuba atholakalayo nesikhathi sokubuyiselwa okuzenzakalelayo okulandelayo. Lawa maphuzu ayindlela yokudlala yendawo yale phrojekthi kuphela futhi awahlobene ne-akhawunti yamaphuzu esemthethweni kaMcDonald's.

Dala i-oda langempela ngokuphepha

Uma udonsa ikhadi olithandayo noma wenza isicelo sokuthenga ngqo, iphrojekthi ayidali i-oda ngokushesha, kodwa isebenzisa inqubo yokuqinisekisa okwesibili okuphoqelekile:

1. Khetha isitolo nokudla noma iphakethe elingathengwa.
2. Biza i-calculate-price ukuze uthole intengo yesikhathi sangempela nezindlela zokuthatha ezitholakalayo.
3. Khiqiza i-quoteToken esetshenziswa kanye enesikhathi sokusebenza semizuzu engu-15.
4. Bonisa umphumela wokuhlola intengo kumsebenzisi bese ulinda ukuqinisekiswa okucacile.
5. Iseva iqinisekisa i-Token, bese ibiza i-create-order ngemva kokuphumelela.
6. Buyisa inombolo ye-oda nesixhumanisi sokukhokha.

I-create-order kufanele iphathe i-takeWayCode, futhi leli nani kufanele lithathwe ku-takeWayList ebuyiswe yi-calculate-price. Ungaliqageli noma ulisuse, ngaphandle kwalokho isicelo singabuyisa u-400.

Ukudla okukodwa kungase kungakwazi uku-odwa ngokuzimele kumenyu ehloliwe. I-README ibhala ukuthi izinto zokudla ezingu-96 zine-canWithOrder engu-false, ngakho amakhadi ephakethe angu-18 yiwona angase a-odwe kakhulu. Amakhadi okudla okukodwa asengasetshenziswa kumaqoqo nasekuboniseni ulwazi lokudla.

Okuhlangenwe nakho komugqa womyalo ngaphandle kwe-Token

Indawo yokugcina iza ne-sample_data, ngakho ungakwazi ukuqinisekisa lonke uhlelo lokudonsa amakhadi ungaxhunyiwe ku-inthanethi. Ngena kufolda yezikripti:

cd skills/mcd-gacha-collector/scripts

Hlola ukusatshalaliswa kwechibi lamakhadi

python gacha.py sample_data --inspect --budget 50

Lo myalo uhlola ichibi lamakhadi esampula ngaphansi kwesabelomali esingu-50 yuan, futhi ulungele ukuqinisekisa amakhadi angadonswa nokusatshalaliswa kokungavamile.

Donsa kanye nokudonsa okuyishumi

python gacha.py sample_data --draw 1 --budget 50 --ledger /tmp/ledger.json
python gacha.py sample_data --draw 10 --budget 50 --ledger /tmp/ledger.json

I---ledger icacisa indlela ye-ledger. Ngokusebenzisa ifayela elifanayo ngokulandelana, ungagcina amaphuzu namarekhodi amaqoqo.

Buka izibalo zamaqoqo

python gacha.py sample_data --collection --ledger /tmp/ledger.json

Lingisa ukusatshalaliswa kwamathuba

python gacha.py sample_data --simulate 100 --seed 42

Ukusabalalisa okuhlosiwe kwesampula ngu-SSR 5%, UR 20%, R 75%. I---seed 42 yenza imiphumela engahleliwe iphinde ikhiqizeke, okulungele ukulungisa amaphutha nokuqhathanisa imiphumela ngaphambi nangemva kokushintsha i-algorithm.

Yenza i-HTML yekhadi

python render.py /tmp/ledger.json --out /tmp/cards.html --last-draw 1

Isinikezeli siguqula idatha yeleja ibe yi-HTML eyifayela elilodwa, ukuze kubukwe amakhadi akhishwe okokugcina esipheqululini.

Ukuzenzela ukuhlola

python selftest.py

Umqondo oyinhloko uqukethe ukuhlola okuzimele okungu-105. Ngemva kokushintsha ukubalwa kwamaphuzu, amaphuzu, izinga lokutholakala kancane noma ikhodi yokukhipha amakhadi, kufanele uphinde usebenzise lo myalo.

Ukuqonda impilo nezinga lokutholakala kancane

Impilo yomkhiqizo ngamunye ibalwa ngokuhlanganisa izici ezintathu ezinesisindo:

• Ubuningi bamaprotheni bungu-35%.
• Inkomba yamakhalori aphansi ingu-40%.
• Inkomba yesodium ephansi ingu-25%.

Izinga lokutholakala kancane lisebenzisa ukuhlukanisa okuhlobene, kusetshenziswa imingcele yamaphesenti esethwe ngokusabalala kwedatha yamanje. Lokhu kugwema ukuthi isitolo esithile singabi nawo nhlobo ama-SSR ngenxa yokuntuleka kwemikhiqizo enamaphuzu aphezulu njengekhofi. Iqoqo lamakhadi elahlolwa ku-README, ngemva kokusebenzisa amaphesenti okusabalala, lathola amakhadi angu-3 e-SSR, angu-7 e-UR kanye nangu-13 e-R.

Amasethi okudla awasebenzisi ngokuqondile izisindo zemikhiqizo eyodwa, ngoba inani lawo lamakhalori liphakeme ngokwemvelo. Iphrojekthi isebenzisa i-algorithm ehlukile yamasethi: ubuningi bamaprotheni bungu-40%, ubuningi besodium bungu-35%, kanti umthwalo wamakhalori ungu-25%, futhi imingcele yamaphesenti okusabalala ibalwa ngokuhlukile.

Impilo nezinga lokutholakala kancane kusetshenziselwa ukuqhathanisa kokuzijabulisa kuphela; akusona iseluleko sezokwelapha, somsoco noma sobunye ubungcweti.

Ukuthuthukisa okuthuthukile namasu okulungisa amaphutha

Ukuhlaziya kahle idatha yomsoco ye-MCP

list-nutrition-foods ingabuyisa data njengochungechunge lwe-CSV, hhayi njengohlu lwe-JSON. Iphrojekthi inohlelo olukhethekile lokuhlaziya olubizwa ngokuthi parse_nutrition_csv. Lapho welula iklayenti, ungacabangi ngokuqondile ukuthi ihlale iwuhlu.

Ukuvula imvilophu ye-structuredContent

Idatha yebhizinisi ye-MCP ingatholakala ku-structuredContent.data, futhi amanani ayo nawo angaba isichazamazwi, uhlu noma uchungechunge. Lapho uphatha impendulo, qala uvule imvilophu yangaphandle, bese uthumela impendulo kumhlaziyi ofanele ngokohlobo lwayo lwangempela.

Ukugcina ama-algorithm e-frontend ne-backend ehambisana

Inguqulo yewebhu isebenzisa ukubala ngesikhathi sokwakha: tools/gen_web_data.py ikhiqiza idatha, bese web/build.py ifaka idatha ekhasini eliyifayela elilodwa. I-frontend ayiphindi yakhe enye i-algorithm yokubala amaphuzu, ngaleyo ndlela kunciphe amathuba okungahambisani kwemiphumela phakathi kwezinhlangothi zombili.

Ngemva kokushintsha ukukhiqizwa kwedatha, ukuhlanganiswa kwezinkambu noma ithempulethi yekhasi lewebhu, ungasebenzisa tools/parity_check.py. I-repository iqopha ukuhlola okungu-99 kokuhambisana, okuhlanganisa i-HTML, i-JavaScript, ukusabalala kweqoqo lamakhadi nokuhambisana kwezinkambu.

Ukuqinisekisa ukuhlaziywa kwezitolo ku-backend

I-backend iku-server directory, futhi isebenzisa umtapo ojwayelekile we-Python ekutholeni indawo nge-IP, iklayenti le-MCP kanye nesixhumi esibonakalayo sokufaka ama-oda. Umqondo wokuhlaziya izitolo unokuqinisekisa okungu-15 okwengeziwe. Kuhlanganiswe nokuzihlola kwe-Skill okungu-105 kanye nokuhlola kwewebhu okungu-99, iphrojekthi iqopha iziqinisekiso ezingu-219 sezizonke.

Ungathembeli ezixhumanisini zesitolo

Ngokwemiphumela yokuhlola eku-README, izinkambu zezixhumanisi ezibuyiswa yi-query-nearby-stores zingaba zonke ngu-0.0. Isu elithembeke kakhulu ukuthola idolobha nge-IP kuqala, bese usesha usebenzisa idolobha namagama angukhiye endawo, futhi uhlele ngokusebenzisa i-distance ebuyiswa yisixhumi esibonakalayo.

Ukuhlukanisa ikhadi lokubonisa nekhadi elinga-odwa

Ukudla okuvela kumenyu akusho ukuthi konke kungathengwa ngokuzimele. Lapho wakha umsebenzi wokuthenga, kufanele uhlole isimo sokuthi kungafakwa i-oda esibuyiswa yi-MCP, futhi uqondise umsebenzisi kuqala kumasethi asemthethweni; ungacabangi ukuthi ikhadi lingadala i-oda ngenxa yokuthi liyaboniswa kuphela.

Ukuvikela i-Token nemisebenzi yama-oda

• Faka i-Token ngokusebenzisa okuguquguqukayo kwemvelo kuqala.
• Ungabhali i-Token yangempela ku-mcp-config.example.json.
• Ngaphambi kokufaka i-oda, hlola kabusha intengo njalo.
• Qinisekisa i-quoteToken esetshenziswa kanye nesikhathi sayo sokusebenza.
• Biza i-create-order kuphela ngemva kokuqinisekiswa okucacile ngumsebenzisi.
• Lapho uhlola ama-oda, phatha lokhu ngokucophelela futhi uqinisekise ukuthi kufanele akhanselwe ngokushesha yini.

Obani abafanele ukukusebenzisa

• Abathuthukisi abafuna ukufunda inqubo yokubiza i-MCP yakwaMcDonald's.
• Abasebenzi abafuna ukuthuthukisa i-WorkBuddy Agent Skill.
• Abasebenzisi abathanda ukuqoqwa kwamakhadi, ukukhipha amakhadi kanye nokugcwalisa i-albhamu yamakhadi.
• Abasebenzi basehhovisi noma abafundi abafuna ukuqhathanisa amakhalori nokwakheka komsoco wokudla ngendlela elula.
• Onjiniyela abadinga ukucwaninga ukuhlaziywa kwedatha ye-MCP, ukuqinisekisa intengo kanye nenqubo ephephile yokufaka ama-oda.

Isifinyezo

UMqoqi Wamakhadi kaMcDonald's uhlanganisa ukusesha imenyu, ukuqhathanisa umsoco, ukuqoqwa kwamakhadi nokufaka i-oda langempela kube yinqubo ephelele. Lapho uqala ukuyisebenzisa, ungabuka kuqala ukusebenzelana ekhasini eliku-inthanethi, bese usebenzisa idatha yesibonelo ukuze wenze ukuhlola komugqa womyalo; ngemva kokuqinisekisa ukuthi umqondo usebenza kahle, lungisa i-McDonald's MCP Token bese ufaka i-Skill ku-WorkBuddy.

Ekusetshenzisweni kwangempela, naka ngokukhethekile umehluko wamafomethi edatha, imikhawulo yesixhumi esibonakalayo sezitolo, isimo sokuthi amasethi angafakwa yini ku-oda kanye nendlela yokuqinisekisa okwesibili. Zonke izintengo, izimo zokutholakala kanye nolwazi lokudla kufanele kuqinisekiswe ngemiphumela yesikhathi sangempela evela eziteshini ezisemthethweni zakwaMcDonald's.