[{"data":1,"prerenderedAt":555},["ShallowReactive",2],{"project-47899844":3},{"id":4,"name":5,"line":6,"line_source":7,"title":8,"url":9,"domain":10,"maker":11,"posted_at":12,"year":13,"points":14,"comments":15,"rank_in_year":16,"shown_in_year":17,"top":18,"tier":19,"category":20,"subcategory":21,"format":22,"keywords":23,"reception":29,"sentiment":30,"stance_mix":31,"scored_comments":35,"aspects":36,"strip":46,"post_text":47,"what":48,"said":80,"maker_in_thread":173,"shape":185,"grid":189,"attempts":518,"same_week":519,"maker_shows":42,"maker_hidden":548,"reader_version":549,"card":550},"47899844","WUPHF","A Karpathy-style LLM wiki your agents maintain (Markdown and Git)","title","Show HN: A Karpathy-style LLM wiki your agents maintain (Markdown and Git)","https:\u002F\u002Fgithub.com\u002Fnex-crm\u002Fwuphf","github.com","najmuzzaman",1777107233,2026,260,114,183,4366,5,"legend","productivity","agent knowledge wiki","cli",[24,25,26,27,28],"llm wiki","markdown git","bm25 search","ai agents","knowledge base","curious",-0.01,{"positive":32,"neutral":33,"mixed":34,"critical":32},10,26,2,48,{"idea":37,"design":40,"privacy":43,"docs":45},{"positive":38,"critical":39},4,6,{"positive":41,"critical":42},3,1,{"positive":44,"critical":44},0,{"positive":44,"critical":34},[42,42,41,44,44,41,18,41,41,18,42,44,41,44,42,41,44,42,41,41,41,41,41,41,41,41,44,41,41,41,18,41,18,18,44,18,38,44,41,42],"I shipped a wiki layer for AI agents that uses markdown + git as the source of truth, with a bleve (BM25) + SQLite index on top. No vector or graph db yet.\n\nIt runs locally in ~\u002F.wuphf\u002Fwiki\u002F and you can git clone it out if you want to take your knowledge with you.\n\nThe shape is the one Karpathy has been circling for a while: an LLM-native knowledge substrate that agents both read from and write into, so context compounds across sessions rather than getting re-pasted every morning. Most implementations of that idea land on Postgres, pgvector, Neo4j, Kafka, and a dashboard.\n\nI wanted to go back ",{"problem":49,"audience":55,"pricing":55,"motivation":56,"time_to_build":55,"stack":60,"openness":77,"stage":78,"team_size":79},{"value":50,"source":51,"evidence":52},"Agents lose context across sessions; need a durable shared knowledge store they read and write","maker",[53],{"id":54},"post",null,{"value":57,"source":51,"evidence":58},"Maker wanted to see how far markdown and git could go before adding heavier infrastructure",[59],{"id":54},[61,65,69,73],{"value":62,"source":51,"evidence":63},"Markdown",[64],{"id":54},{"value":66,"source":51,"evidence":67},"Git",[68],{"id":54},{"value":70,"source":51,"evidence":71},"Bleve (BM25)",[72],{"id":54},{"value":74,"source":51,"evidence":75},"SQLite",[76],{"id":54},"open_source","launched","team",{"praise":81,"objections":90,"requests":120,"questions":135,"alternatives":154},[82],{"value":83,"source":84,"evidence":85},"BM25-first over vector DBs is a good call","commenter",[86],{"id":87,"by":88,"text":89},"47900083","Unsponsoredio","love the bm25-first call over vector dbs. most teams jump to vectors before measuring anything",[91,98,109],{"value":92,"source":84,"evidence":93},"LLM-maintained markdown collections may degrade quality without human curation",[94],{"id":95,"by":96,"text":97},"47900375","stingraycharles","The few scientific studies out there actually show a degradation of output quality when these markdown collections are fully LLM maintained (opposed to an increase when they’re…",{"value":99,"source":84,"evidence":100},"Name and product seem like satire or redundant",[101,105],{"id":102,"by":103,"text":104},"47901280","arikrahman","I thought this was parody at first as well for a redundant useless product as it was named after the redundant useless product of the same name from The Office (Wuphf.com)",{"id":106,"by":107,"text":108},"47901447","410298","I no longer can tell if this is satire or not: https:\u002F\u002Fwuphf.team\u002F It looks like gastown. Using AI is like children playing with a dollhouse. If it is satire, well done.",{"value":110,"source":84,"evidence":111},"Agent-generated content gets confidently wrong over time",[112,116],{"id":113,"by":114,"text":115},"47901491","Abby_101","The \"garbage facts in, garbage briefs out\" caveat is the part I'd want stress tested. In my own LLM features the context that decays fastest is what agents wrote without a human…",{"id":117,"by":118,"text":119},"47902027","ryanshrott","One practical approach that works is separating the capture layer from the promotion layer. Agents can draft freely, but anything that gets promoted to trusted status needs a…",[121,128],{"value":122,"source":84,"evidence":123},"Configurable wiki directory",[124],{"id":125,"by":126,"text":127},"47900347","hansmayer","Couldn't you instruct your LLM to make the starting dir configurable?",{"value":129,"source":84,"evidence":130},"Support for arbitrary OpenAI-compatible endpoints like DeepSeek",[131],{"id":132,"by":133,"text":134},"47901185","LanMeng-LM","Would be great if the provider layer supported arbitrary OpenAI-compatible endpoints — DeepSeek, for example. Any plans?",[136,143,147],{"value":137,"source":84,"evidence":138},"Does it support multiple humans?",[139],{"id":140,"by":141,"text":142},"47908137","psteinweber","I couldn't find out on the fly: does it support multiple humans? A lot of these tools (paperclip) don't, but for our small, agent-supported team, that would be the feature we're…",{"value":144,"source":84,"evidence":145},"Does promotion require human review or can agents self-promote?",[146],{"id":113,"by":114,"text":115},{"value":148,"source":84,"evidence":149},"How to deploy securely for sensitive documents?",[150],{"id":151,"by":152,"text":153},"47900554","sails","How do you anticipate teams deploying this? I’m wary of GitHub for sensitive business documents, and wonder what an easy secure agent friendly deployment looks like. Cloudflare…",[155,162],{"value":156,"source":84,"evidence":157},"Obsidian vault with a plugin",[158],{"id":159,"by":160,"text":161},"47899966","davedigerati","why not an Obsidian vault with a plugin?",{"value":163,"source":84,"evidence":164},"Other LLM wiki projects such as commonplace and Shoggoth.db",[165,169],{"id":166,"by":167,"text":168},"47901360","zby","Reviewed: https:\u002F\u002Fzby.github.io\u002Fcommonplace\u002Fagent-memory-systems\u002Freviews\u002Fwuphf\u002F It is a third llm wiki on front page in 24 hours! Obviously it is a hot topic. I have my own horse…",{"id":170,"by":171,"text":172},"47900421","GistNoesis","The space of self building artefacts is interesting and is booming now because recent LLM versions are becoming good at it fast (in particular if they are of the \"coding\" kind).…",{"comments":174,"answered":39,"key":175},12,[176,179,182],{"id":177,"text":178},"47902029","thanks for saying this. one more reason people are hating on SaaS so much is that the UI is dry and for many, unusable. on the other hand, the cool AI agents have UI which is fun…",{"id":180,"text":181},"47901976","alright sir\u002Fma'am\u002Fneither. we built an AI-native CRM backed by HubSpot founder Dharmesh Shah last year before this, had revenue, iterated to focus on context graph infra which…",{"id":183,"text":184},"47902075","love the \"Borrowable Ideas\" section. would suggest to definitely borrow them. full disclosure: we started as a context infra company (nex.ai) from long long before Karpathy even…",{"commenters":186,"max_depth":39,"maker_replies":174,"first_comment_mins":187,"lifespan_hours":188},81,15,112,[190,194,198,201,202,205,209,213,216,219,222,224,227,230,233,236,240,244,247,250,254,257,259,262,264,267,270,274,276,279,282,285,288,291,293,294,295,298,301,303,305,307,310,313,315,317,320,323,325,329,332,334,338,339,343,346,349,351,354,356,358,362,365,367,370,373,375,378,380,384,386,388,391,393,396,399,401,404,407,411,413,416,421,426,428,431,435,438,443,447,449,451,453,455,458,462,466,470,473,476,479,482,485,489,491,493,495,498,500,502,506,509,511,513,516],[191,44,42,192,193,44],"47900197","project\u002Fcritical\u002Fidea","portly",[195,44,44,196,197,44],"47900373","commenter\u002Fneutral\u002Fnone","_zoltan_",[199,44,44,196,200,44],"47902420","Bridged7756",[95,44,42,192,96,44],[203,44,44,196,204,44],"47900417","criley2",[206,44,41,207,208,44],"47902795","project\u002Fneutral\u002Fidea","saadn92",[210,44,44,211,212,44],"47900405","other\u002Fneutral\u002Fnone","mplappert",[214,44,44,211,215,44],"47900508","simsla",[217,44,44,211,218,44],"47900888","mohamedkoubaa",[220,44,44,211,221,44],"47901069","danielbln",[223,44,44,211,218,44],"47902386",[225,44,44,211,226,44],"47903721","rafaelmn",[228,44,44,196,229,44],"47904021","skybrian",[231,44,44,211,232,44],"47902631","4b11b4",[234,44,44,211,235,44],"47903741","nicbou",[237,44,41,238,239,44],"47900428","project\u002Fmixed\u002Fidea","frocodillo",[241,44,41,242,243,44],"47900756","maker\u002Fneutral\u002Fidea","frrandias",[245,44,44,211,246,44],"47900760","bushido",[248,44,44,196,249,44],"47900992","jstummbillig",[251,44,18,252,253,44],"47900471","project\u002Fpositive\u002Fidea","0123456789ABCDE",[255,44,44,211,256,44],"47901022","emsign",[102,44,41,258,103,44],"project\u002Fneutral\u002Fdesign",[260,44,44,211,261,44],"47901403","big_man_ting",[263,44,44,211,200,44],"47902451",[265,44,41,207,266,44],"47902771","anuramat",[268,44,18,252,269,44],"47904191","adamsmark",[271,44,41,272,273,44],"47904997","project\u002Fneutral\u002Fprivacy","johntash",[275,44,44,211,269,44],"47956804",[277,42,42,192,278,44],"47900222","batoga",[280,42,44,211,281,44],"47900533","girvo",[283,42,44,211,284,44],"47900681","ting0",[286,42,44,211,287,44],"47901721","rglover",[289,42,44,196,290,44],"47901913","hmokiguess",[177,42,44,292,11,42],"maker\u002Fpositive\u002Fdesign",[180,42,44,242,11,42],[166,34,41,207,167,44],[296,34,41,297,243,44],"47901981","project\u002Fneutral\u002Fnone",[299,34,44,196,300,44],"47902057","Myrmornis",[302,34,44,196,167,44],"47902136",[183,34,44,304,11,42],"maker\u002Fpositive\u002Fidea",[306,34,44,211,232,44],"47902645",[308,34,42,192,309,44],"47903901","SOLAR_FIELDS",[311,41,44,211,312,44],"47899990","mellosouls",[314,38,44,211,290,44],"47901936",[316,18,41,207,273,44],"47905066",[318,39,44,211,319,44],"47903196","mncharity",[321,39,44,211,322,44],"47908792","jermolene",[170,324,44,211,171,44],7,[326,327,42,192,328,44],"47900272",8,"dataviz1000",[330,327,42,192,331,44],"47901410","iterateoften",[333,327,44,211,328,44],"47901500",[335,336,44,211,337,44],"47900949",9,"RivoLink",[113,32,41,207,114,44],[340,341,41,238,342,44],"47903600",11,"renan_warmling",[344,174,41,207,345,44],"47899922","dhruv3006",[347,174,41,207,348,44],"47900173","left-struck",[350,174,44,211,345,44],"47900313",[352,174,44,211,353,44],"47902173","kaoD",[355,174,44,211,345,44],"47908212",[140,357,41,207,141,44],13,[359,360,44,211,361,44],"47919663",14,"gbram",[363,187,41,258,364,44],"47900176","armcat",[366,187,41,258,253,44],"47901295",[368,369,44,211,171,44],"47901587",16,[371,369,44,211,372,44],"47903983","genewitch",[151,374,41,272,152,44],17,[376,374,41,377,243,44],"47900774","maker\u002Fneutral\u002Fprivacy",[379,374,41,258,152,44],"47901017",[381,382,44,211,383,44],"47900185",18,"imafish",[385,382,44,196,126,44],"47900338",[387,382,41,297,243,44],"47900558",[389,382,41,207,390,44],"47900622","jbjbjbjb",[392,382,44,242,11,42],"47902896",[394,382,44,211,395,44],"47900637","psanchez",[397,382,44,196,398,44],"47901061","dominotw",[400,382,44,196,395,44],"47902280",[402,382,44,196,403,44],"47900675","stavros",[405,382,44,406,11,42],"47902879","maker\u002Fneutral\u002Fnone",[408,409,44,211,410,44],"47906730",19,"manfre",[132,412,41,258,133,44],20,[414,412,44,415,11,42],"47902999","maker\u002Fneutral\u002Fdesign",[417,418,18,419,420,44],"47911793",21,"project\u002Fpositive\u002Fdesign","michelhabib",[422,423,42,424,425,44],"47900201",22,"project\u002Fcritical\u002Fdocs","souravroy78",[427,423,44,242,11,42],"47905402",[429,423,18,430,425,44],"47907386","maker\u002Fpositive\u002Fnone",[432,433,42,424,434,44],"47901627",23,"Invictus0",[436,433,44,437,11,42],"47902563","maker\u002Fneutral\u002Fdocs",[439,440,18,441,442,44],"47899960",24,"project\u002Fpositive\u002Fnone","goodra7174",[444,445,44,211,446,44],"47901808",25,"smadam9",[106,33,42,448,107,44],"project\u002Fcritical\u002Fdesign",[450,33,44,406,11,42],"47905349",[87,452,18,252,88,44],27,[117,454,41,207,118,44],28,[456,454,18,252,457,44],"47915713","drewbatcheller",[459,460,18,419,461,44],"47900238",29,"vlady_nyz",[463,464,44,211,465,44],"47901854",30,"mehhehheehh",[467,468,44,211,469,44],"47900002",31,"hyperionultra",[471,468,44,211,472,44],"47900014","spiderfarmer",[474,468,44,211,475,44],"47900070","wiseowise",[477,468,44,211,478,44],"47903083","kid64",[480,468,44,211,481,44],"47900045","William_BB",[483,468,44,211,484,44],"47900073","mirekrusin",[486,487,18,441,488,44],"47906667",32,"samarthv",[159,490,41,207,160,44],33,[492,490,18,419,160,44],"47899972",[494,490,41,406,243,44],"47900475",[496,490,44,196,497,44],"47900097","tomtomistaken",[499,490,44,207,11,42],"47901876",[501,490,44,211,478,44],"47903076",[503,504,41,258,505,44],"47900034",34,"jimmypk",[507,504,41,258,508,44],"47900497","tomjwxf",[125,510,41,258,126,44],35,[512,510,44,258,11,42],"47905379",[514,510,42,515,126,44],"47908818","maker\u002Fcritical\u002Fnone",[517,510,44,406,11,42],"47927895",[],[520,525,529,533,538,543],{"id":521,"name":522,"points":523,"category":524},"47920787","dirac",393,"dev_tools",{"id":526,"name":527,"points":528,"category":524},"47874647","Honker",327,{"id":530,"name":531,"points":532,"category":20},"47882697","Tolaria",318,{"id":534,"name":535,"points":536,"category":537},"47886517","How LLMs Work",245,"learning_science",{"id":539,"name":540,"points":541,"category":542},"47937380","Auto-Architecture",241,"hardware",{"id":544,"name":545,"points":546,"category":547},"47876071","Turning a Gaussian Splat into a videogame",236,"games",false,"v2",{"id":551,"name":5,"line":6,"ai":548,"cat":20,"year":13,"points":14,"comments":15,"rec":29,"s":30,"top":18,"kw":552,"maker":11,"domain":10,"tier":19,"attempts":44,"strip":553,"url":9,"mix":554,"praise":83,"objection":92,"sub":21},47899844,[24,25,26,27,28],"1130035335103013013333333303335355054031",[32,33,34,32],1794036699516]