16:[[47616116,{"by":"Acacian","descendants":1,"id":47616116,"kids":[47616145,47636866],"score":1,"time":1775145113,"title":"Show HN: I tested 11 AI frameworks for basic security – none passed","type":"story","url":"https://acacian.github.io/aegis/playground/"}],[47542749,{"by":"Acacian","descendants":0,"id":47542749,"kids":[47542793,47542794],"score":3,"time":1774620005,"title":"Show HN: Aegis – Security framework for AI agents","type":"story","url":"https://acacian.github.io/aegis/playground/"}],[47501426,{"by":"dot_treo","descendants":500,"id":47501426,"kids":[47502858,47502785,47509479,47502402,47513932,47502085,47507089,47502459,47502350,47504118,47502034,47502920,47502769,47502493,47502548,47503196,47504121,47503114,47502619,47502297,47505844,47502002,47502008,47503842,47503580,47502296,47502301,47502319,47503373,47507836,47517825,47502346,47502839,47502260,47510471,47502955,47510350,47513787,47506099,47506165,47595397,47504191,47512083,47502412,47503065,47502237,47503031,47510246,47511845,47502586,47502568,47502109,47504933,47506200,47503270,47506350,47502731,47514873,47502529,47506858,47510199,47503343,47530192,47508422,47502447,47522157,47502309,47504387,47502440,47502380,47515626,47511696,47510301,47506618,47506658,47507548,47512863,47504732,47502604,47508281,47502454,47504102,47506974,47502549,47508540,47511296,47505215,47503725,47515544,47501928,47505713,47502355,47503685,47508315,47502925,47502906,47502353,47502980,47507593,47501993,47502063,47501432,47511188,47506076,47519380,47503702,47505214,47511907,47503106,47501856,47519747,47529617,47507545,47502272,47507550,47510846,47512178,47511149,47504719,47516215,47512165,47502434,47522817,47502798,47515479,47511671,47502856,47508055,47518203,47512080,47516771,47526751,47503281,47518256,47501658,47506347,47513129,47529741,47506356,47506701,47502542,47505992,47503590,47508120,47515393],"score":938,"text":"About an hour ago new versions have been deployed to PyPI.

I was just setting up a new project, and things behaved weirdly. My laptop ran out of RAM, it looked like a forkbomb was running.

I've investigated, and found that a base64 encoded blob has been added to proxy_server.py.

It writes and decodes another file which it then runs.

I'm in the process of reporting this upstream, but wanted to give everyone here a headsup.

It is also reported in this issue:\nhttps://github.com/BerriAI/litellm/issues/24512","time":1774353989,"title":"Tell HN: Litellm 1.82.7 and 1.82.8 on PyPI are compromised","type":"story","url":"https://github.com/BerriAI/litellm/issues/24512"}],[47520220,{"by":"jawiggins","descendants":60,"id":47520220,"kids":[47528314,47525453,47598854,47525855,47524587,47528440,47527060,47527007,47530431,47527135,47530037,47531010,47525213,47524594,47526092,47525242,47622871,47530263,47529759,47524726,47560717,47532271,47529597,47557361,47525020,47530136],"score":88,"text":"$17","time":1774458621,"title":"Show HN: Optio – Orchestrate AI coding agents in K8s to go from ticket to PR","type":"story","url":"https://github.com/jonwiggins/optio"}],[47524704,{"by":"marciopuga","descendants":49,"id":47524704,"kids":[47525608,47525180,47527529,47524986,47526195,47526218,47527705,47525837,47525819,47619757,47528725,47527198,47528700,47526399,47530410,47531811,47524963,47527483,47534902,47527687,47527967,47689345,47560045,47575419,47526849,47530761,47524717,47560099,47529590,47527435,47557369,47561604,47534303,47528412,47526377],"score":157,"time":1774481353,"title":"Show HN: A plain-text cognitive architecture for Claude Code","type":"story","url":"https://lab.puga.com.br/cog/"}],[47521157,{"by":"louiereederson","descendants":222,"id":47521157,"kids":[47522974,47523456,47522858,47523261,47525140,47529586,47523414,47522840,47524609,47525215,47525838,47522807,47523926,47523035,47523030,47536816,47524000,47529679,47528162,47522895,47523144,47525985,47541517,47524010,47525834,47527979,47524701,47525228,47523798,47526445,47522691,47523295,47524033,47545477,47528183,47529631,47525543,47523712,47529448,47527563,47527534,47524240,47526917,47529503,47527138,47525692,47524405,47524526,47529854,47526782,47529424,47522925,47529316,47527352,47527131,47527919,47523617,47524298,47528796,47528252,47525567,47530783,47529274,47523924,47524672,47528682,47522927,47529028,47529305,47525844,47527182,47526340,47528073,47525818,47523238,47523752,47523134,47521158],"score":337,"time":1774462600,"title":"90% of Claude-linked output going to GitHub repos w <2 stars","type":"story","url":"https://www.claudescode.dev/?window=since_launch"}],[47495510,{"by":"latand6","descendants":29,"id":47495510,"kids":[47496557,47501877,47497552,47530661,47509300,47496900,47525767,47509319,47536748,47497473,47500238,47552610,47497379,47496752,47498046,47555066,47506635,47516520,47496711,47497373],"score":10,"text":"$18","time":1774302192,"title":"Ask HN: Is using AI tooling for a PhD literature review dishonest?","type":"story"}],[47501690,{"by":"klueinc","descendants":23,"id":47501690,"kids":[47502399,47503342,47514033,47539657,47530875,47502638,47502476,47502257,47507372,47502128,47502916,47504411,47506115,47515556,47506180,47525896,47527339,47512193,47519731,47501819,47506991,47520086,47520081,47526890],"score":14,"text":"I currently have a claude pro monthly subscription ($20) which I use for coding. It's been useful but I'm fatigued from optimising my work around it's session limits. There are so many choices and providers out there today but hard to get a good signal about what's good. I'm not looking for another Opus-level model but something reliable enough that it can follow TDD well.","time":1774355501,"title":"Ask HN: $50 monthly budget, which coding models would you recommend now?","type":"story"}],[47490244,{"by":"sayYayToLife","descendants":8,"id":47490244,"kids":[47490563,47493066,47493561,47491233,47500613,47502450,47515214,47496042,47518422,47501806,47512688,47493780,47490456],"score":7,"text":"As I use open claw I am concerned about prompt injection more than bad code. However I think I'm irrationally paranoid. I'm small fries I'm a single individual, someone actively trying to hack me is exploiting a multi-million dollar zero day and they're not doing that on me.

If we're thinking about accidentally deleting everything through a terminal command, I've yet to see this actually occur.

Deductively I can see all of the worst case scenarios with open claw. Inductively I've never seen it actually happen.

I find it a bit irrational to pretend that open claw is a genuine security risk.

The moment I see on Hacker News that someone got prompt injected, I think I'll be concerned. Until then I would need almost a lottery like chance to get hacked as the first person through prompt injection.","time":1774277011,"title":"Ask HN: If there has been no prompt injection, is it safe?","type":"story"}]]

user:	Acacian
created:	March 21, 2026
karma:	11
about:	Building Aegis — open-source security framework for AI agents. https://github.com/Acacian/aegis

1.	Show HN: I tested 11 AI frameworks for basic security – none passed(acacian.github.io) 1 points by Acacian 91 days ago \| 1 comment
2.	1 points by Acacian 94 days ago \| discuss
3.	Show HN: Aegis – Security framework for AI agents(acacian.github.io) 3 points by Acacian 98 days ago \| 0 comments