[{"data":1,"prerenderedAt":184},["ShallowReactive",2],{"project-43364640":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":27,"sentiment":28,"stance_mix":29,"scored_comments":33,"aspects":34,"strip":38,"post_text":39,"what":40,"said":54,"maker_in_thread":123,"shape":126,"grid":130,"attempts":145,"same_week":146,"maker_shows":176,"maker_hidden":177,"reader_version":178,"card":179},"43364640","LLM-docs","LLM-docs, software documentation intended for consumption by LLMs","title","Show HN: LLM-docs, software documentation intended for consumption by LLMs","https:\u002F\u002Fgithub.com\u002FDicklesworthstone\u002Fllm-docs","github.com","eigenvalue",1741971860,2025,28,7,1710,3837,45,"shown","ai","llm documentation","other",[21,24,25,26],"rag","documentation distillation","llm context","quiet",null,{"positive":30,"neutral":31,"mixed":32,"critical":30},0,2,1,3,{"idea":35,"execution":36,"pricing":37},{"positive":30,"critical":30},{"positive":30,"critical":30},{"positive":30,"critical":30},[30,33,30,33,30,30,33],"I was inspired by a recent tweet by Andrej Karpathy, as well as my own experience copying and pasting a bunch of html docs into Claude yesterday and bemoaning how long-winded and poorly formatted it was.\n\nI’m trying to decide if I should make it into a full-fledged service and completely automate the process of generating the distilled documentation.\n\nProblem is that it would cost a lot in API tokens and wouldn’t generate any revenue (plus it would have to be updated as documentation changes significantly). Maybe Anthropic wants to fund it as a public good? Let me know!",{"problem":41,"audience":28,"pricing":28,"motivation":47,"time_to_build":28,"stack":51,"openness":52,"stage":53,"team_size":52},{"value":42,"source":43,"evidence":44},"Software docs are long-winded and poorly formatted for pasting into LLMs","maker",[45],{"id":46},"post",{"value":48,"source":43,"evidence":49},"Inspired by a Karpathy tweet and frustration pasting HTML docs into Claude",[50],{"id":46},[],"unknown","prototype",{"praise":55,"objections":64,"requests":82,"questions":95,"alternatives":111},[56],{"value":57,"source":58,"evidence":59},"Idea of machine-readable docs is valuable and fertile ground","commenter",[60],{"id":61,"by":62,"text":63},"43369057","darkteflon","It’s a cool idea. I’ve wasted a lot of time over the past few months futzing around with beautifulsoup, Playwright and others I forget, or cloning entire repos and trying to…",[65,74],{"value":66,"source":58,"evidence":67},"Distilled docs may not be what LLMs need; retrieval is the harder problem",[68,71],{"id":69,"by":62,"text":70},"43369334","Yeah, agree that one could \u002F should definitely put it through evals if it were going to be a product. There’s quite a bit to a good evals program, though. I think the main…",{"id":72,"by":11,"text":73},"43369146","I feel like humans don’t have great intuitions around the ideal format for an LLM. But an LLM probably does have a good sense for what would work well for itself. And also I…",{"value":75,"source":43,"evidence":76},"Token cost to generate docs is high and updates are costly",[77,78],{"id":46},{"id":79,"by":80,"text":81},"43385990","eternityforest","Why does it need to cost a lot of tokens, and why can't it incrementally update? How do the VS code integrated AI helpers do it?",[83,87,91],{"value":84,"source":58,"evidence":85},"Plain text or markdown versions of docs rather than LLM summaries",[86],{"id":61,"by":62,"text":63},{"value":88,"source":58,"evidence":89},"Evaluate distilled docs empirically with task-based evals",[90],{"id":72,"by":11,"text":73},{"value":92,"source":58,"evidence":93},"Incremental updates instead of full regeneration",[94],{"id":79,"by":80,"text":81},[96,100,104],{"value":97,"source":43,"evidence":98},"Should the maker turn it into a full service?",[99],{"id":46},{"value":101,"source":58,"evidence":102},"How do VS Code AI helpers index docs?",[103],{"id":79,"by":80,"text":81},{"value":105,"source":58,"evidence":106},"Guidance on writing docs for LLM consumption",[107],{"id":108,"by":109,"text":110},"43371130","rikroots","Oh! Recently I had the experience of working with someone who was using LLMs to build something using my JS canvas library. The code the LLM was producing for this person was ...…",[112,119],{"value":113,"source":58,"evidence":114},"Dagster's docs chatbot works well for RAG",[115],{"id":116,"by":117,"text":118},"43369207","Noumenon72","Dagster's docs must be LLM-readable somehow because their LLM that lets you ask questions about their docs is the best RAG experience I've had yet.",{"value":120,"source":58,"evidence":121},"Docling and unstructured for document conversion",[122],{"id":61,"by":62,"text":63},{"comments":32,"answered":30,"key":124},[125],{"id":72,"text":73},{"commenters":127,"max_depth":33,"maker_replies":32,"first_comment_mins":128,"lifespan_hours":129},5,496,62,[131,133,135,137,139,141,143],[108,30,30,132,109,30],"commenter\u002Fneutral\u002Fdocs",[61,32,33,134,62,30],"project\u002Fmixed\u002Fidea",[72,32,30,136,11,32],"project\u002Fneutral\u002Fidea",[69,32,33,138,62,30],"project\u002Fneutral\u002Fexecution",[116,32,30,140,117,30],"other\u002Fneutral\u002Fnone",[142,32,30,140,62,30],"43369364",[79,31,33,144,80,30],"project\u002Fneutral\u002Fpricing",[],[147,152,157,162,167,172],{"id":148,"name":149,"points":150,"category":151},"43358914","Nash",536,"productivity",{"id":153,"name":154,"points":155,"category":156},"43401855","BlueMigrate",524,"infrastructure",{"id":158,"name":159,"points":160,"category":161},"43373242","my-yt",319,"consumer_social",{"id":163,"name":164,"points":165,"category":166},"43404548","\"Git who\"",252,"dev_tools",{"id":168,"name":169,"points":170,"category":171},"43398081","MinesweeperPro",242,"games",{"id":173,"name":174,"points":175,"category":166},"43377897","Cascii",238,15,false,"v2",{"id":180,"name":5,"line":6,"ai":177,"cat":20,"year":13,"points":14,"comments":15,"rec":27,"s":28,"top":18,"kw":181,"maker":11,"domain":10,"tier":19,"attempts":30,"strip":182,"url":9,"mix":183,"praise":57,"objection":66,"sub":21},43364640,[21,24,25,26],"0303003",[30,31,32,30],1794036671231]