[{"data":1,"prerenderedAt":140},["ShallowReactive",2],{"project-40671686":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":31,"aspects":32,"strip":34,"post_text":36,"what":37,"said":59,"maker_in_thread":91,"shape":93,"grid":96,"attempts":101,"same_week":102,"maker_shows":132,"maker_hidden":133,"reader_version":134,"card":135},"40671686","promptfoo red teaming","Automated red teaming for your LLM app","title","Show HN: Automated red teaming for your LLM app","https:\u002F\u002Fwww.promptfoo.dev\u002Fdocs\u002Fred-team\u002F","promptfoo.dev","typpo",1718296159,2024,23,2,1790,3515,51,"shown","ai","llm red teaming","library",[21,24,25,26],"prompt injection","llm security testing","jailbreak testing","quiet",null,{"positive":30,"neutral":31,"mixed":30,"critical":30},0,1,{"idea":33},{"positive":30,"critical":30},[30,35],3,"Hi HN,\n\nI built this open-source LLM red teaming tool based on my experience scaling LLMs at a big co to millions of users... and seeing all the bad things people did.\n\nHow it works:\n\n- Uses an unaligned model to create toxic inputs\n\n- Runs these inputs through your app using different techniques: raw, prompt injection, and a chain-of-thought jailbreak that tries to re-frame the request to trick the LLM.\n\n- Probes a bunch of other failure cases (e.g. will your customer support bot recommend a competitor? Does it think it can process a refund when it can't?  Will it leak your user's address?)\n\n",{"problem":38,"audience":44,"pricing":28,"motivation":48,"time_to_build":28,"stack":52,"openness":57,"stage":58,"team_size":58},{"value":39,"source":40,"evidence":41},"Finding failures in LLM apps such as toxic outputs, prompt injection, jailbreaks, competitor recommendations, false refunds and data leaks","maker",[42],{"id":43},"post",{"value":45,"source":40,"evidence":46},"Businesses shipping LLM applications",[47],{"id":43},{"value":49,"source":40,"evidence":50},"Maker scaled LLMs at a large company and saw bad user behavior",[51],{"id":43},[53],{"value":54,"source":40,"evidence":55},"promptfoo",[56],{"id":43},"open_source","unknown",{"praise":60,"objections":73,"requests":74,"questions":82,"alternatives":87},[61,69],{"value":62,"source":63,"evidence":64},"promptfoo is a pleasant, flexible eval tool","commenter",[65],{"id":66,"by":67,"text":68},"40672026","danenania","I haven't yet tried this red teaming tool, but I recently started using promptfoo to build out an evals pipeline for Plandex, a terminal-based AI coding tool I'm building[1].…",{"value":70,"source":63,"evidence":71},"Good foundation to build red teaming on top of",[72],{"id":66,"by":67,"text":68},[],[75],{"value":76,"source":63,"evidence":77},"Support dynamic prompts and providers for continuous evaluation in production",[78],{"id":79,"by":80,"text":81},"40673973","Oras","Can this be dynamic on prompts and providers? I’m thinking of continuous evaluation for LLM in production, where after each call, a webhook will send the input\u002Foutput to evaluate.",[83],{"value":84,"source":63,"evidence":85},"Can the tool run dynamically on prompts and providers?",[86],{"id":79,"by":80,"text":81},[88],{"value":54,"source":63,"evidence":89},[90],{"id":66,"by":67,"text":68},{"comments":30,"answered":30,"key":92},[],{"commenters":15,"max_depth":31,"maker_replies":30,"first_comment_mins":94,"lifespan_hours":95},32,4,[97,99],[66,30,30,98,67,30],"other\u002Fpositive\u002Fnone",[79,31,35,100,80,30],"project\u002Fneutral\u002Fidea",[],[103,108,113,118,123,127],{"id":104,"name":105,"points":106,"category":107},"40694254","MUREX Robotics network switch",666,"hardware",{"id":109,"name":110,"points":111,"category":112},"40717398","Radius",351,"consumer_social",{"id":114,"name":115,"points":116,"category":117},"40669337","Shpool",323,"dev_tools",{"id":119,"name":120,"points":121,"category":122},"40719782","Billard",306,"creative",{"id":124,"name":125,"points":126,"category":117},"40710154","Tokencost",268,{"id":128,"name":129,"points":130,"category":131},"40722317","RTranslator",236,"languages",10,false,"v2",{"id":136,"name":5,"line":6,"ai":133,"cat":20,"year":13,"points":14,"comments":15,"rec":27,"s":28,"top":18,"kw":137,"maker":11,"domain":10,"tier":19,"attempts":30,"strip":138,"url":9,"mix":139,"praise":62,"objection":-1,"sub":21},40671686,[21,24,25,26],"03",[30,31,30,30],1794036656296]