Skip to main content
Glama

Tecton MCP Server

Official
by tecton-ai
examples_rift.parquet34.2 kB
PAR1���ML��wpExample of Entity. An adV$Content IDZ!(Auction ID)VBA userp0the platformXV-TThe keyword describing:c �0this ad is beDplaced alongside.F6\XFeatureTable. Precomput3�d embeddings pushed into Tecton.H�J�vLE6LAttribut:��I�6I$realtime_f �_view. C �s-<hsine similarity between a pZ and:6�>h�CfpdX6JsC6�4RequestSource. s$ for query9q input.S6G��>J�6�.|6W�qNr6�:��6�$test. TestA�aU'1_X_U�_9�' M@ which takes in r)i data (':B')%6�J( uo-TputsQ6�stream:�a�count�a�$mpressions%�aq�_�:2Aggregaa@�G�!44past 1 minuteY6��\\5 \s\6]�. y�givenIA�a ad>��A50-ad pair over9hourW6��WW24Zs�[J[72 [A6��mad clickQv!:AZ�B7JEJ6��J18�N N1+:��+.���a%0K��V<J5O�OO1:�SS1?:jS��lV僁�>_���G:�&E^ed:2�Q�:K?6KZ�.,2NA�f6Cbatch:�(How many di��@ct advertisementsM ha� en shown�ul�uweek]6j��a a�6aN�TI ��5 run�� pipel1 ,mode to turn�-�a-�}, an array��� s,+\n create metrics based oHat 2.`6�=Lis� qs extrac� frombA\yT'sQ�1 field.Vbd Numb� fi��Z>�Y�Wheth�Q \ l�A�ain� !-, 'bitcoin'.c6�pyarrow_)�8config. Declare�J  t!\connects!�4your Iceberg t : BA�j 6�iD�nA? .� 8synchronous, onEPing� onA,ad-e� eve: with�p 16 meta� � n);aI�+fig��backfillA�offptrainC genera� .k�����S info.t6o�&Servi� A: provi� q���a�l)�predi!�if�u will��e�d.^�$ merchant[6��y transa�am� , us!cU � ed schemaj|."FT�higiCai�81 day average.Z6�Z� ���^h6^�edE��-�1� �I�>� �FBoolean6indica� if%pf�.M6���:��!$100.N�Q��0far apart twoe$s' home lo�� are>R�Distance�� kilometer�Q�bWi6���e�$age calcul%�Q�aMƉ�a !?stamp�Y:��User'sXinA�s.96�j��:=:�v�i{Ey.�tude � +>� Y}~�.nd�2�:u Z�U22�is��-\j6G� w�X map,u,%struct)ʩ]a�lex�+types.,6n=OutputV��strB.>~0 two-dimen�al �.B: Fr �� w�float64 �.�6�ZzRea:� D%��s`o�c =.%9�M�.�6�J�d�L_�L _is_��_than_���amt�B�� ("_mean_1dS tinuous5� ����A�_ag���8gj�X0date_of_birth1�p��fO>@ depends oB-B u56t�I�� loginU s Dime.m69R[Aerrehave>E�of al� e ag�2 funws avail� �^�D6q*hSumR=��he �B� � e. M�of�I:�IC5 :JR�f�MinimumJ�%F��jHax�H^bdPop�< �dar�vi�O:�n~:� =5S��^T�� variz��PbX ��TebTApproxim[\50th percentile (median)��>�Y�First 2JeR�:IL�F���E��f��"� R�>�9 ��G:"RQ�N+( 3"� ed):�**�JBblMost reA_ 10J�sAjL6F �:n �P:= RP&�]_2� made �lyf_"(non-fraudul�2Wper)���a"= window:K],�^30@ _����� \:/ RlT.� ���tc$ total1serieE hs, up�d�ryAU�sjks��f��Gdayf�F3Hsf�� a>]cjf��Hday>5�ЦG �z6��h �! .�e0s at specific>~a�8daily>Q�d!X>KxaqI�"� dayf�T30Ws�XBX9 Xb����� �� r8��~�)j86f=jQ2is_ [V.R<�<-:�@-":�b���A3t , enteredE�ignup.4:�t"�"LatH?Er�"�:@ .8on�V9:b�ct cq 4issuer derived&��.)n�.>��e�.2�[��.�peroJ.8i� ,18 years oldj��[[:Z]�jMeach pu[sStegory:M�>�fnk�g:jbK.qw&�split byF�.d6�a+-֞�f�j�B�f�F-��J J��q�B%w=�Ee=n� toN� �~ �>�}��lFl:� N#D�؉z�o M'�Y7�V>�&��)�su?2��ev Q6" ]:�N�ve 2.� :�=n�I:����F�:��:� u� end.�N�N�I6��"+R^6�M�>4:���:XN�Me.�� ���.ufY T(:�)�f:t 2� ���?-�� demo:�FileCos. �uRM%3 c,{�Parqueql��!S3�6�.b j.]re J. l� S�l orma"�*� aI�y �.66�6� � .}>s.l6: Push �� fi2�d-�c �H#��.�s_ &' ��!:] %�  -|%�.�j� o�.$!� �u�:g�S�  j?u�C rela�#�-*$Ev .:6B ^ ]. F� dete�s �:(>>j\!�#TB�#0 Item/an ?/U O.te!6�0�/s�+:G/E�:IUnique1s�articl�te�$::�N.u�L��.�C:cN�AJ�:"7 � of�"s,� t$rd�on!F �f� ��c�*rF��&+ z�1o)1, 3,f�a*�"3 A�i ()F]a�%e�":-=��_��`f`! :�v�m��:�).� a�:2.>I���p6z>� Z�-��b���mak��2mendaE�;6tv�ga���s{':VSnowflak�W�8.�acc�C4FR��G:M���! �:�^ �N�R�FY2�:J�b���D-�%�gam:5y9nt>�ޥ��)�2�:U�1 N�-!�F�:H":%� he mob�>:-$0 prod�"2:=:��##(a TIMESTAMPH":;*L �N� elapZ�1 currBimif�a 's ��6OZ�"�m(co%� v�"payload�3�-~>� (fetc7HaM�!��Z�.).�6�R.��* -�b--�c,i�6 ��&shistory.6�-�#ay objectE�follow�@$ure: {'v _1':"�r��%, 32': ...}:*]%Gs�U�T� a JSON �R6N�-q!p. G� ID��Tim�'ofYV9�*!e� fNv@>6�䞂�6BZ�TBexpenses�Xros�'@ nkB" �8�5Rd*\ !�Plaid.�s APIq:j�'!RvV:?=*A'�.%s�$v �$N�s,eNa@[30, 60, 90, 120]��,�6(�Example of test. Testing the 'user_query_embedding_similarity' feature which takes in request data ('query_embedding') and a precomputed feature ('user_embedding') as inputsYExample of Aggregate. An array of sums of transactions for every day over the past 7 days����/  !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abTUcdefghijklmnopqrstuvwxyz{|}~y��k���������������������������������������������&�U 5text����U&�M&6(�Example of test. Testing the 'user_query_embedding_similarity' feature which takes in request data ('query_embedding') and a precomputed feature ('user_embedding') as inputsYExample of Aggregate. An array of sums of transactions for every day over the past 7 days,��ȿL~��t�from tecton import Entity .types dField, String, Int64 ad = ;<( name='ad',(join_keys=[A('ad_id' A)] '<description='An @@owner='demo-user@ �.ai a\tags={'release': 'producK0'} ) contentN�" ",Z� � �R�C dID �� �a �R� b� ��A d��) !�N�'ads_-F��� '< of the platform T�T-T1�keyworV�-�K # |F�:K'qR  The tI� bing� H4 this ad is be@placed alongside. ��%�2R�`Timestamp, Array, Float64m�.(�"TeatureTable, Attribute +datetime- delta ads.e��i�s4ad schema = [e�REi� Y�',�6#0ad_embedding' �( �)) ] .# s = .�6�Ks-�� =[ad��f)s= �5 (�3J ', d�}=%|� � T�_f��=.�%,(online=True ff: ttl=H%�(days=10 gI��� PrecomputA�d 5?8s pushed into T�R�� ) ���r=�-�U�I<�QZ.�B�Ng�s Ij���9�nP*B -Ċ.N�"!�e�1�^��Aj2�1� real�_-7_viewB�I:���)�A Bs _t��s.�9n"�:�H6@c 2 @R��_sourc�125,>�MH,mode='pythonMp�js@cosine_similarityU^-0)�H d�x��.U"C�Rs�� �, between a pZ~ and:!�Y�." ) def_.K�(.>b):%-�(numpy as np. .linalg) orm #~Bi$(a: np.nda�]b. s# Handle)"Pase where one or both�o, do not have:��.� if alNK �(return -1.0 � 4np.dot(a, b)/(� (a)*b)) /resul| {}{ ["B "] =J(!�y�["6"],2U[".�"]).as�H('f�'I� � � |VnRequestSe�,�}�V���� r �_&+ r('query9>F�]< =6�( L=6[)�� 6>K��> B)B������%hye��>9�.�ZI�!����:�t��is no2�a�9������J�-�[":��� �-� math* ��$.on_demand2��:w]AQ#!�j' # Testwyu fX'^$ which tak� n1 I a F�) #J � ( wխ')�jinputs�� testej�(m-� = {B: [1.0,  0.0]ɜ%&~:690919��actual = Cf8.�runqڱa>+=6y�M#� � arisoniNexpec� = 0.5 �ssertEP .isclose( �[JS],D ) R} stream2�`, DatabricksClusterConfig xeamProcessingMode, Aggregate, \�ionLea�Edg2R� �KInt32}(rQ�T�&� data_[ � impr�on<�_ � B>,�D @ 22).� =Ra �=[:���� andaW np5j_%S%�6�@.CONTINUOUS, # en� , low latency Ming%�m�2B 1�(e�_column=%�('clicke %�), func!�='count')*_window*z$minutes=1)}�m~m5m-^[&zQ Fals:R-6_start_T=Q(((2022, 5, 1 �v�Io~.T j%lEOY� for a>/a-�ion_li�_edge=1�6�D.LATEST_EVENT_TIMEi :�_E_�si ��� df =:v[���q�', ]_*�]].copy(� df['Q 2  ).abs("! int B df m�1%��60�AN�e+,� & � ��ad�#"����������� mg��'2�m4�jYFJ3�m hour�k�n n24�mU/RG�o o72o-u������batchh dule��:��-����9LWPgiven� � "v� � A�_2�� Euu5� 5�:� # ReV �< to id"�u df.r((��In: id'}�:# Add=X < with��0stant value 12��9��� �q�V��QV�ͯ�q���P>r! �n�n�nm��j�j��� 64�� �g�k�d�lBaS* n�@�@�@ad %+= a2 ��^/ �F2F��>EE�2@6��D�" �" ����������Z��j��� � � � � � � � b R����2�Y����w�� �� �� &� I� )_ �(�W,�I'�Fd#V B� $TriggerTyp' 6�$ B6& �(�N�:�)� �%���vW ed.ns� key-�&groupe��events uZ�y#- H Ingest API. # # SeA documenta�@: # https://docs.-ai/X-the-iJ0ion-api/#crea%-a-� -� -� - -a-push- �:4-_) = �=�/� ="29 s_fv�" `=2-X7�^L� n% �^� 3, 1&�bAr� manual_trE�0_backfill_endF,36,F�%3�%�HVH>� -�A�A� b� �=>y.MANUAL)�a�r�-r2 q�('&� #'e�z] ) ��MFM"mi�m)Y "� Bool�v6*l$Q����d J�f�=1m:a3M�r?3 -key�Zd-%4�<�<�<� ��1F6wa>6$=Q,.unfiltered(MQ1�=�]I��:>:0alert_email="J�.��NRH%�� m$�������������^V^E���>gA�"+�V_�pM�%�X���p�-Co�.&6�E* ��qK���2�s_� A>� #� !, pd7)F22x-*q v4.select_range(*�FP.MATERIALIZATION_STAR� -E/�� 6), ���LEND0�$1c2cJ�J�q)�%n�incre� l"xs. /.*��&$ Qv=?1���,^�(i ads'��HowT y di�"(ct advertis�s aE� ha(en shown in7 last week�� -n���e�"�&e\�Y�*aQ_B3_7d:,,} xt5# G  by C�+s. _ids��6�.] by(*�).agg({3s(: 'nunique'4}).reset_index"}s{m$ ( SQL output)df.#= [ ��B'�7�5� gjipd.��u (� xt.q�) -!i�(microsecond!T&��V�trans�4\ ,m�M/��c�~��.�0�1P 32,d �<N��5!����� T�4 �  runsa` pipen2��AI,^&� �q&:2an& -�words,-n t@e metrics based oa�at 2. @� �%o�JHr%� ��-V U\�BY1 �j "�list"6�3)����� "num!s=B �X>5-p_A�ai�itcoinA� @�Fl�mfM�`-.�N�#v� e� 0-�5I:k�bN/*j"P# Spli�'2�"M�Q�-���[' E����!�en&�"].str.s`(' �. vCalcu�%\ stata.'a6v^len���!�.(^B@apply(lambda x: 'MX'a�x�>�| Aya��)ir�A& ��e[[6�:��, �aA2� F�] $N�<Hiv �(Kinesis6�(0� Da�8Parti%C�, Push @��# F �G�q � .2P.�N)u=��� ! B.Nkpyarrowy !�F��:xt��Declar72J1 t�lconnect��4your Iceberg tH( @N:��sup�s�_dingX m�6� i \ �( 4t& �py +.catalo"�3load_ �+ _!i = k_ads"eVc�*ur���7{"1":"} �any!fi3 param�4��a#_%%_�{&� "2� =2�(�z , **6�irtb�, ".%|(�"O,A�E�! )g�f avail%��U ![� if �!�&94B.&0 & V�Kend((�Q[*", ">=", *lQ) �Fx� �v<FuN s%#-E ��an)�sca�? tbl. �j�s� =. IM�s-��( .to_e�()� Use�pY ���<Be =.#i�A��?F. [ Ij>�,m{  �:�'"��;jz 0 { ^=�V����"%�'V5B ,', � mobil� "p*�>��("y"` � "# cast('�?�� C�iF } +�~Ab,Brzn&v���r��mmpaigvBZLv�4partner_domain��H^���*6 �0�� f/v�ad_7.A^�ad�play_�A��de�Ń1N�v�1PN:vCis_pwa�) < #�:(nd � s boolea�M���X^?) Ś*" )Q#  _utc`0�Be�v �0:� ��Q" "�2�.1f+�i�..�,m� ] q��~f<&� x1 /- �=.� ��4.]0""+ A :NZ0,synchronous,� in���ion�"��/< a�"� meta�. � �  a�j�4figuA����inL gene� 15""-'U :_H !�3.* stag8D�i3N��E�)��E�.  ���b�bqI0!�M5jd1%� ,- �J I`20�%Q%� info. ~��~�9V$K� Servicl��sU*%7�9�^"�32QR!� �])��6U.�Mv�BJWe��-B�_7A.�#!�R!�GctrM_s)v&�G-�:N J3� .�'A:� provi�8��*Dlpredi � will#+ a�L�FF9.�Li,I�j�LF �ng_�7dJC.q%S.� mB�: N�:#N���], ) �Re�aNpGRjLfraud<-�F�K ��b�Q�:'��< �lLFlLmerch 1= �.�b�!v�(%��Sv� ) �V�ReB� �C)�2�InF"@#g �\%�z_{ag:_amount_-@!�j'jBCamc �)] .l&b>2^;C�$&�'.R �Xis_higher_than_average'�)]��C.Y �,~ m'��=ZNH�Io2Q�Nv �B ϡD�e 1 day �.�  <v (.*��� �_me�0��z[A1_1EinuouYis@G else�Ge��{�:22-� '] >.�} �6nMq2���m#6�^F! A�� !R'"�M�X�X}I,# An example� an on-jC�de�s�y�a�k+� . ��KJ��y�y $100�lN]m<��~�100} Z�V��U��cR]N� home_loc�q5�02C F>us�Lsam�$6Y4as two separat2�D. �M�s #��Fbv�"sendegU" "recipi!id"����F���F.�_�8_map({> id"::�}��.-x _km.; �.�"$far a%^�s' !� 1�s!4Penviron� s=['P,-core-1.1.0':��J_to  ance(-w[�I1xmC��Mr�N� O = >Y['lat�E "][F] 5�B .- ERongH!n� =� ��@) # In kilometers ĕ,F!�:P} <�d �d e͟��0date_of_birth1�E2n�M.���vMt �A*; � ޢ + 2�mP� "���'s ag��� s:�Qge�HN�.(�!@1z4Qh�O�,='.amiso�%u[=�]�&�(tz`=QiFdob�U�'U'�&�aR�WtsZ-#�-2m *u@b!�,: td.days} ��@�@6� �B�s�z1��2N+ %10 ��&0 ����.��PZ_��\�{ -�M�2��`��.� � A�!['s��+��.� ���_Y�q?N�.���1�I����YU��R!����J-n?3�2�h,.N��_R�R�WZ�)1*2 ib�.�!"�*fefɄr ucj!0Mapf=�%}�o�X =6� i��("s[b: ",h)"m� 28two_dimensional!ay"kU(%) E)J>si� s uc�St � �  w �w]"��� -nJ64.-$2/�p 2��&"L-�ma�8F=�CFIZN� -�CzCf��"*[",�s=��_I�],� s"��&|Qlex�a�%_odfv��� # T. map G%�B� =[1!�IB.["newa 0Y ]@&.ya�*.zR�.�R!6�R#�%["t"]�$.�secF�2=.�26�25<i��M� �P  J>:Fpy�0R�;RFF2�>2 6} �� :>ds.�vB�s�1j*py�T��"gv��qXz:Vy~.VamR�)T� s�*�.mark.bz͂"daily�,,KU= ��(100, 25��  10, �.J:%M�N�V(��*�UM(�NR�cA{^7: `./.V W!�: �}� [V =24)SZ�& Wu�Jp=Jiz�� .B �-��1��J.V�':9).rWN=- �# ���:�k�)2q�Ph%�VZ6�.RU�(�.{28%Q'2O)(1992-12-05')>.`�@: '2021-05-14T00:.000+ '>nag]^ Y:iY2�I(29� { ��10387 �-�q=4 gr��ft��>�>�[f\>:&�5�<%�Y~TidjI9:�DF@�V�o�:( �5�M�.ME�6��2�.7 Z�29Q 2l65��.Y �N�I�Mk��aYQk/Q6q!,�"T�/Z�6b$2�aJ?*7ta� 5A=���a��2J� &O ih logi+%_7�!��[n-30..]G".O��%%��)�2,�!�2�=6" ɘ�#�2AJ�(J]77 �o~�Yv�! �. 'User %O %& over%�"�$�m= .Il ) Q;6]�(-s92�]�Bio��]U�N {..�KZ\Uz)�(approx_perc!�l H firs9�=)(�inc ) ���R.�nA0I� �0_A2� �}��Z^# Areh�ve%��of al�&e =% 9%"�3,2�@��YJ�):1 2Z�^? S*n�@MD.�&Q>m�� 4u^"am�M8mG1="sum �A�F�^�D �.R "Sum!p= s%� pY@LaU��B�meat��M#of����:�����1�C'&:�����=�/�inimumJ����ax��1�Max����F�$stddev_pop���Pop�=;1R[ dard devi�"����B�-sam�5S�%������var��Z� vari�����F����J�����b�BZ (&e =0.5,0ci�=10�Q��A� im&50th v (median)��BZ (.�D�F�  2J����+ ����L0 FR ������9������mz1^�" E�"� ���Q�w \"�1��V���!���J�-���#���.�; �[K�RH��6!g�nk Ls.�F0N}��jOw4R,6�F9�]]E��$VU�Q&`�U�v����.��o.T� eam��U�6ZjM4ZMFB-�5�Ŗ�J�"�Y.M'e;A8.LP (I c�Id)� �23Q'(M}-�=Zd*l).�n�j=�pyspark�T�aO!)}a ]] ;V�O)9q�Ҟ. �1"�ߑ�i�6I�":�+Bq�� H �KzK'h�? llow'?ef�J a sli;!y w�^. 3col�K�He%  NJ�s� (;��I�1+Fm6��� # D �how fr ntly ; job0�>�d�P_�M�1�@ stor`"N��HV�YoQ.Qi2V(��%�f���f � ͖ pr@�Droy H+FSet!=� ) accid�`l� $�(;ges�~ downA{�A���42q Most�0�10v��:a�_re{.Rads�@~_]0&�OA^% �rstrI1� df 1V�*R?fc6Sv&� ���RT.�q���#}5fJF�9-#�;m2 # I Ru�c�)numbe`> non-��ulAw.��E �,a 1min, 5min�51h��� �"freshnes�@these�s.<1�V�~�~M3��xT -B2�rA=B��i�mx�q q30�+��n�{�Xi�� f��2�������b�Nu�=���� &\ made�!ly���.�.� E�.- �7 �LH�f.'�EN9 s_z3�:8j>[2��B1��.( ��5U��*�= � �!�!.!��F"se�Al~D  's.�� s�����0 �0 �0 M .1 'su <A�j% i�U�R0e9 &�5..eng�7=��~�f3�?ff* b��8�4)4rh�595�g56�A.r��(�� ��'�� F� ^2�F�1'tccs8 totalT seri^ f�L L�p�*d�Ory  &��2��p �-N��u5� A�V�V�V�V�V�V�V�V�V�V�V�Vsql�E�VR�@o��.��u(&� j�:� ��� ,"(.� � ?����V�,@22�K*�)=.�G�]�J��� J�.RintervaF���}-l������ʄn��o9"-��������-=j�-(at specificu4"V ~��5��a!Z .%c2��)Ny",66:�T���������͉{��������mJ��&��r�is_%ORf�� �� �j���kB��-�� 2QI� �L��K:Kh)�_h2?��5J [[2;=K2G �V F��5��bE�79"� � =���B_..�vm�6� �# Note��W�signup�, henc (ol�j*��v�17,�v{N�!65�)��F��!�=L$tZx�% b� &�`�-2 &�!N"I.#G*��], .�2�H(BA6�B[[J��sr�! ��N] KV�!ڣ�8��������*�������z�>�8Z&j �� �� �� �.c[^t c�14issuer derived6�K2)c�Iv! a_a_ a&�"� .F.b .� N�)F� 'cc_num:� �#d�w $I"8H�get�9 digit3�[N��." z. !�a [0].�TS'4�@ Visa��'5Ma��ar�'6 DiscF "1w�na('o+��id��EPj6E/2� ]] ��O2O��= �[E�{f@�%���!3� *�B ~B # For�*fzv*�0>Z��`c��aqs'indicate�Wwhe!�� wa� ad �as �� 2J�Ҕ,)3)]N �� � �:JM�r�v ��~ <J&N3�� F� W-�!����#0T�#=�� a� 18 years� �|Z3BE"�e�q A v��%�PJB J � # Merg inMns�i�� a�BM MRm> e�� !,>ʱT�� how='inne�)#*�vG>V��z Ps)H�*um^l�&2��((5[']�{to�E!$dob'])).dt_U( > (18*365)F� �S�vd[(order final �F�vk%�� �� �']] JRmAi;qn�}��]{"�.I�� �n��>�oolRDreduce CATEGORIES�X%q'ga #5h�'�  kids_pet�C'shoppQpo groc�'neXfoo�o?l_e� �l"\ perswRcar�h��h_fit�'+ misc.wstravel.� !] �le"���D) r� f'{��gory}_f �� ���s��t�*Ǟ��BY1!,%� atte�.e `*�s` 4z�qs..aSI~(�z,l, r: l + r,>n=~)$is��_�ie�ZG4&�~�$each purchɫ.�R $hic.y, e.g.�9 much #�&���InY{ "6z"!&�B30�[�`�I ��[�`1-*F*� #!8Q14�U  ao f 28^s|�acquiva�*6z an b�)lsF�  2+���s(PIVOT clausecFɒ���|J�1 6�� �6 H�A 7,"-�2�5���Yt<~ byFt���Řs7ݩ. g_1es_by`{�T6�dfnFFdf��# CP�e�"c?}�,e,q#.Aica�1}��Pdf[�q��� df�TlyIm� row:[+%]hrow['tB=qe�gMT, �axis=#)2� F�F��5�'] + 2�V��>!�^] �� � �����m�� ~� ��^����� " .}(fb�'���}��n��rw�ooB��������VFVB��'��r�=��RW'B�6j��2W� !�x�&���?݈�p�Y;. 1�T R�� g�� �,�Lky�َ��2��"� .� VU �6�l "custom�� ion". `f�`� a�9s��̃ #UNu�� i��30* �% j͋.1� k�i��cu- aNžmor~Nfojz�Ԇst/|�,s/framework/q�  /R�.htmlj�j& �$�^$�29�%��%���4&s Ra*-%���F��9��2 $e many2�F �= �3toN�B.�5 �M�Z� >j*%2�T�� �wF� PwsY�"q>zd6� %�,����� >?�id�kJ�B< [:D�idFB��!�E�r�g"0��Nè.o��F�5���(�� - 1 m&+�)�������|�>|�^�, �� ]] ��#��k imeW @SM0� >�!X+��� ?�"Qa _�,RrA�� "��I��+m�Q��*.�"D|`��r.� sum( A��7� �!c.�B&GN�1�6 NNU� ",)A�,���zNU>,6vW&� >F�- Y.�B _sizbv�:�2t An aeof-i2�T{v3�x�B��|i�]2�*� _!�2��|� +b� , '%�a� �V�[��>$ j)�H�H� �>(2�/���� ��*���� 3~&.�zHm�H�H��D]��52�)�f� [.xq6&� ]] ���.����7�Q�"I I�jH�R&=;�.) �q��2%{<>x#9�plek y� �l� ���� �A5zK .�Bnstead�4�Z�w�n� , it~s�0's built-in `R�`._�� ��R���6EH�"��9*A .��I�:"� NMHRQ()INF.=H���-6�} �} &} >�*���E� ��� ��i>�, 9� ]] y��2�q�E7�M< ���`�_ ?�|�G��I>� �����?r?is_��en}2 �xa0�"m1�?�a*am�cost-:JerF�nce��2D" �2q�A�& j��v i2)�i30je�9��AdfId� \ # Ge_�'�@eek (Monday=0, Su 6��� � .ś[.?9 �].*$of�(.isin([5, 6&p�iO�� P(hU�J�, *� )6�9�|��}��}�) � ��#&�N�uR���Bs.� 3 3v��i/�i/� 3� 3� 36 3� . Q�j��_o�&�2��FQ..2Jb�.�>F@F �Forma��t�+�3� yyyy-MM-d$�)��6�spdZ�(.d�rf�$('%Y-%m-%d.C."v�>d/�7�Q%%�N9>;�j;:#� DSrɟ�.  2tB�~ʝ�� .1^&B_2;Ns �&�&�&�&�&&.�� B ?*�6&� 9�:�a� 2d Mya.Z �$�$#�Qim!WNMuWv��a2Y LprZ ]Fj��_R�File>�.g�m7a����ff�� %-s _cs�=d&2$uri='s3://q�@i.public/tutorial%ud_:�/ ers/ bfile_�/at�rqu"�+B�B�7� �� ) V�3�,���ڡm�y  � f�3-Tj�Z,.|�)�*64.Om0V1�&�� .�ЕU])���.�1�j�!�R| �2*؟}("ʊ,f��"���n+g(r%�n�͊)"~�����'doub�� �?`AsA�.;�!I� 9T9 M�.%_la&�v�`_wb<v;5�:j%�)�A�>�onA8�"9�&.�t\` I� =���� #I�*t.�.�I>= W��y (-Ud*��" �6��6ntsMp.?=. �r2�f�F,�68�. ���:V�j7S뙉V{ &�F�~�WB����R�Y=Ss&�bv&�v� ?��ԙc�6�]f�0b:-�.=' K -@�c�8�6� reh%"�*U�!. l�F�"Bj2C��ro:�"���*��=Jk�Mc��)'*��� )�V�V � �R�m�-���.�_ �P1� �&�(:'Rki7fj"fb RZFJ�V�~Ѣ1R�N�A^�hzR�%R�4���Xan.[FkJ � de �!-�Irz���f<MÖ�@1 94�!\�Ų A�6a �.E��i~%�J8mj)I�:!��6N���@ #��� Z�_v2\ �p�(9��~�:v2',32k�@�@�=\�EU"`�y�V��}�=�;%�has_gk3�A eNew421Z�cRm���s�(:�67�JdU~B�.-Q�B%��9] )G�mZw al_f�3FXIV)E>2�.��x-C #$~) �Z�2�R$R�&�bF3�3%~n)W� c�;U-�r�6�fI5]a8!8~4$�� Z�*� �.g�8|���V���)́�or�(y�clE 8.= -V��s&����� 2{ Item�n �B����tR�Av�� s{�N!�" V B� j�N� K- ��<V} 2 �0^!W.^������f�recsyJM�rtA,2$2 -'_�i"� � >jWf�.]"UJ�]<$ �kkera �*"�16�v� PM��#6�+"Q "tp.���>�B�F�6���#6 "�* �_1.( P&6�Mi���%'tsX��AIi2�:.R s of3���art0�FCZ��� �m���.��hary.�=ķ���i !$f�� �F . In�4His would�>'= GROUP BY  h5P�imaryk�����?�&�P=����E���!�un�Ÿ6� �=e,i�&E.N=a{f�&I�62@"��c��zGJ�".vMqi� ��4y)�uyp�-.asJ "��=1) m��!�/�x]��@�@� ��d�@�@�P�#�@FQ��yV �. R�)*� ��J 7, 3�] � � >�0&.�e?�^.^I�� �g�d>�vv"�eu,&2|�v.v769m��g(���.�'&\l�Po-MzM21, 3,�3�" !_���B��alZ�j�VVk *��q� � -f> � (-I��>�u���`��o�in��:�j�N�� �^�  �VZN� �&�)(f�s.-��1l2>:M -BMsessio�n_approx_count_articles recommender_feature_service = FeSe,( name='j8',(0description=' I J8 for computing �0 and session �s+mak(uaVs i*=[ H_ Ds,NV�] �onlin5�Irving_enabled=False )�from tecton import BatchSource, SnowflakeConfig f0.types 6�Field, Int64, String, Timestamp, Array gam�user_bj =.r(%a%�"g+s" �2_co�=:� ;` database="VINCE_DEMO_DB B< schema="PUBLICt!P="ONLINE_GAMING_USERS#lurl="https://<your-cluster>. sn-/ -region>.U.com/Twarehou�$COMPUTE_WHt1@_f!`(='TIMESTAMPM"), ) 1UtransacEB�].3 v�d�d(TRANSACTION�k�k�k(# Declare aMJe� the Push i9 input_Ik = [E�e$(E�'AZ_ID', daH=i2)I�.)EVENT_TS*uTJ-AA�E_PLAYE�_ GAME.�e�), ] 2�� � with ae�,1��a2:@ parameters # See%#,API document��%=B�Ce� J�Stream ~,� ,Q�event_so�� =20�%s�P..M�)�=.�)��  �rbt Vv%�y �yRy z{ �}�7s%���A*)�(M�./" -��+,game played %�s" ) �R�Entity.$6�iE,i@U�l = =]'�a�O(join_keys=[i�}� ])�[.�'A dofE�mobile� W0owner='vince@�n.ai Xtags={'release': 'produ��'}�� n� # ZF�0PRODUCT_NAME'�Z�p �0 available in��N��R� real��_ 4_view, Request� Attribute���Q�g$Personaliz��. js.!�_last_A�_M��Ro � =6�(�\QP����UR ) @R�I�2k''NumberAmxminutes elapsed between currente� (comb ���r � payload) eš�!s\ �'s ! E� (fetched Ma�e�  View).''M�i�s=[ u,TB[mT,mode='python A1�=�B5�('�_�_sa[X'� )�] ) def |/install(v�):E%@ dateAQ , (M&pandas d_2 = #.to�['Y'T]).replace(tzinfo=None �!�I�vW%(-I� e2.] W td =F�-No�return {j�: td.I� } Ym�>�e��i&Z ��V�� _Mj1 5.�  C��-��Mk�hmateri���text,Y.JN� 5�MA>delta @FxR�<Aggregate metric9 each��category��aE�h's 30 day purchase history.)� Thisq_ outputsa��  object� �follow�K$structure: K{'� _1':E total|s�%, 32': ...}�5Q=[-�A��#-�F� )�L"~ _sql�Sincre� @l_backfills=True,E� ttl=5�(days=30�)��� dule>'1 &>�_Z� MkZ� PURCHASE� z�2�-Rical_aM&ions(-9. , coE�=Z�()��m� f'''sSELECT �ID)-TO_��('{ g .end�m�}') - INTERVAL '1 MICROSECOND' AS T�2T<CHAR(OBJECT_AGG(��CATEGORY, SUM(QUANTITY)::variant) OVER (PARTIV BY�))p USER9a�FROM {J7}WHERE � _TS <��AND1>=^*startR,30 DAYS)�GROUP�, >�!�T >,E�����6s ���Ԓ�b I� W��n�/ �� �%B� ˩dn�"R5.w "G� ID' "�( �V�n"� � �� H��=M\.�I�n�v)&�%z�~*$ a�_*")��P �B�Q�5J�&off\ �_Y�=Qn$(2023, 1, �#�I " ": "� ion"} �R� ) a6�:A Rx ��)Fi X �� T�$8expenses accros�A�'s bank���Ved�L�-!�� �Plaid T*o� �^I�s=[2 �� �4('PLAID_PAYLOAb) )]H�{Z� ]��_(_sp�.aL _%s_�t'%i-[)eSi�8ange(30,150,30)2 p!� sK �ԅ�Q�� ,E�� a>� json.>� df&� DataF�(6.Y �[:2]).get('.�')i� df['�']\.!  0� _dicz{} ���R1� df_sub =rU>�� E~%4$ i)]|>� = int( t ['amA�'].sum( ��[�!!>h��(output_dict��,�6(�import math from ads.features.on_demand_feature_views.user_query_embedding_similarity import user_query_embedding_similarity # Testing the 'user_query_embedding_similarity' feature which takes in request data ('query_embedding') # and a precomputed feature ('user_embedding') as inputs def test_user_query_embedding_similarity(): request = {'query_embedding': [1.0, 1.0, 0.0]} user_embedding = {'user_embedding': [0.0, 1.0, 1.0]} actual = user_query_embedding_similarity.test_run(request=request, user_embedding=user_embedding) # Float comparison. expected = 0.5 assert math.isclose(actual['cosine_similarity'], expected) � from Personalization.entities import gaming_user from Personalization.data_sources import gaming_transactions_batch from tecton import batch_feature_view, materialization_context, Attribute from tecton.types import String from datetime import timedelta @batch_feature_view( description='''Aggregate metrics for each product category in a user's 30 day purchase history. This feature outputs a Snowflake object with the following structure: {'category_1':'user total purchases in category_1', 'category_2': ...}''', entities=[gaming_user], sources=[gaming_transactions_batch], mode='snowflake_sql', incremental_backfills=True, ttl=timedelta(days=30), batch_schedule=timedelta(days=1), timestamp_field='TIMESTAMP', features=[ Attribute('USER_PURCHASES', String) ] ) def user_categorical_aggregations(gaming_transactions, context=materialization_context()): return f''' SELECT USER_ID, TO_TIMESTAMP('{context.end_time}') - INTERVAL '1 MICROSECOND' AS TIMESTAMP, TO_CHAR(OBJECT_AGG(PRODUCT_CATEGORY, SUM(QUANTITY)::variant) OVER (PARTITION BY USER_ID)) AS USER_PURCHASES FROM {gaming_transactions} WHERE EVENT_TS <TO_TIMESTAMP('{context.end_time}') AND EVENT_TS >= TO_TIMESTAMP('{context.start_time}') - INTERVAL '30 DAYS' GROUP BY USER_ID, PRODUCT_CATEGORY ''' ����@ �@Da�q� H�$��,˲0L�4��8��@PE�$MEQUUY�eY�i �u��y��}���`��(��8�$Y��i��y�h��꺲lۺ���0 ��L�u]׶m����y��y�����|����&μ 5code��� ��&��&�Z6(�import math from ads.features.on_demand_feature_views.user_query_embedding_similarity import user_query_embedding_similarity # Testing the 'user_query_embedding_similarity' feature which takes in request data ('query_embedding') # and a precomputed feature ('user_embedding') as inputs def test_user_query_embedding_similarity(): request = {'query_embedding': [1.0, 1.0, 0.0]} user_embedding = {'user_embedding': [0.0, 1.0, 1.0]} actual = user_query_embedding_similarity.test_run(request=request, user_embedding=user_embedding) # Float comparison. expected = 0.5 assert math.isclose(actual['cosine_similarity'], expected) � from Personalization.entities import gaming_user from Personalization.data_sources import gaming_transactions_batch from tecton import batch_feature_view, materialization_context, Attribute from tecton.types import String from datetime import timedelta @batch_feature_view( description='''Aggregate metrics for each product category in a user's 30 day purchase history. This feature outputs a Snowflake object with the following structure: {'category_1':'user total purchases in category_1', 'category_2': ...}''', entities=[gaming_user], sources=[gaming_transactions_batch], mode='snowflake_sql', incremental_backfills=True, ttl=timedelta(days=30), batch_schedule=timedelta(days=1), timestamp_field='TIMESTAMP', features=[ Attribute('USER_PURCHASES', String) ] ) def user_categorical_aggregations(gaming_transactions, context=materialization_context()): return f''' SELECT USER_ID, TO_TIMESTAMP('{context.end_time}') - INTERVAL '1 MICROSECOND' AS TIMESTAMP, TO_CHAR(OBJECT_AGG(PRODUCT_CATEGORY, SUM(QUANTITY)::variant) OVER (PARTITION BY USER_ID)) AS USER_PURCHASES FROM {gaming_transactions} WHERE EVENT_TS <TO_TIMESTAMP('{context.end_time}') AND EVENT_TS >= TO_TIMESTAMP('{context.start_time}') - INTERVAL '30 DAYS' GROUP BY USER_ID, PRODUCT_CATEGORY ''' ,<5schema %text%L %code%L�,&�U 5text����U&�M&6(�Example of test. Testing the 'user_query_embedding_similarity' feature which takes in request data ('query_embedding') and a precomputed feature ('user_embedding') as inputsYExample of Aggregate. An array of sums of transactions for every day over the past 7 days,&μ 5code��� ��&��&�Z6(�import math from ads.features.on_demand_feature_views.user_query_embedding_similarity import user_query_embedding_similarity # Testing the 'user_query_embedding_similarity' feature which takes in request data ('query_embedding') # and a precomputed feature ('user_embedding') as inputs def test_user_query_embedding_similarity(): request = {'query_embedding': [1.0, 1.0, 0.0]} user_embedding = {'user_embedding': [0.0, 1.0, 1.0]} actual = user_query_embedding_similarity.test_run(request=request, user_embedding=user_embedding) # Float comparison. expected = 0.5 assert math.isclose(actual['cosine_similarity'], expected) � from Personalization.entities import gaming_user from Personalization.data_sources import gaming_transactions_batch from tecton import batch_feature_view, materialization_context, Attribute from tecton.types import String from datetime import timedelta @batch_feature_view( description='''Aggregate metrics for each product category in a user's 30 day purchase history. This feature outputs a Snowflake object with the following structure: {'category_1':'user total purchases in category_1', 'category_2': ...}''', entities=[gaming_user], sources=[gaming_transactions_batch], mode='snowflake_sql', incremental_backfills=True, ttl=timedelta(days=30), batch_schedule=timedelta(days=1), timestamp_field='TIMESTAMP', features=[ Attribute('USER_PURCHASES', String) ] ) def user_categorical_aggregations(gaming_transactions, context=materialization_context()): return f''' SELECT USER_ID, TO_TIMESTAMP('{context.end_time}') - INTERVAL '1 MICROSECOND' AS TIMESTAMP, TO_CHAR(OBJECT_AGG(PRODUCT_CATEGORY, SUM(QUANTITY)::variant) OVER (PARTITION BY USER_ID)) AS USER_PURCHASES FROM {gaming_transactions} WHERE EVENT_TS <TO_TIMESTAMP('{context.end_time}') AND EVENT_TS >= TO_TIMESTAMP('{context.start_time}') - INTERVAL '30 DAYS' GROUP BY USER_ID, PRODUCT_CATEGORY ''' ,�� �&��,pandas�{"index_columns": [], "column_indexes": [], "columns": [{"name": "text", "field_name": "text", "pandas_type": "unicode", "numpy_type": "object", "metadata": null}, {"name": "code", "field_name": "code", "pandas_type": "unicode", "numpy_type": "object", "metadata": null}], "creator": {"library": "pyarrow", "version": "15.0.2"}, "pandas_version": "2.2.3"} ARROW:schema�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 parquet-cpp-arrow version 15.0.2,cPAR1

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/tecton-ai/tecton-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server