Saw your post and thought: maybe i can make someones day so well here i am reading it :D
Big props on the no install / no register etc philosophy. If i would had to make any account i probably would have closed it instantly again xD
So the first thing i STRONGLY recommend, add somewhere a help text (before starting game or small on the side of ingame and ability to disable via options) for the controles.
Im on a Desktop, i started the game and i expected some sort of short info about controles. Yes theres a tutorial - no i didnt play it. I mean lets be honest... tryNSucceed :D
So ye i spend the first 2-3 Stages only spamming spacebar because it didnt came to my mind that maybe its with mouse support (visually it really hard compares to vs-likes that dont support mouse).
So i can tell - ice mage with just massive rapid space spamming works perfectly fine through the whole earth stage. ^^
I find the base look finem the overview texts for the different arch types is well done (even tho it confused me that fire and ice dont have weaknesses?).
The point that at least for me was the most well, unpleasent? , is the size of the play area. I guess you made this to fit easy with smartphone screens, but on a desktop its like not even 1/4 of my screen (and im not on 4k or something) so dunno it feels just alot to small. May fit for a smartphone but for a desktop its just very very limiting while the game takes alot of space for basically nothing.
Also, you definatly should have a "Settings" button in the game pause menu which allows for changing sound levels. Not just a "total sound" bar but at least have Music and Sounds (attack etc) seperated. Because, frankly speaking, the music while for the first like 30 seconds is cool, very fast is dunno it just would fit more to the entry video scene of something than as a constant thing (my pov) - so i wish i would be able to just disable the music and still have the attack/battle sounds. Adjusting both tho would be great anyway and i think with phaser should be quite doable.
A smaller point (visual) is the size of the health/mana bars. Even tho i know they are in the top left, i kinda have to squeece my eyes sometimes to see them. So i would probably just make them bigger.
That all said, i mean i just played solo till the fire stage :) and i clearly had a bit of fun.
I would say its a great start and if you go on and refine it i see a chance that people might pick it up as a nobrainer lets just game something solution :)
Ishikawa : a framework/architecture for automated Attack Surface Mapping & Vulnerability scanning
- golang based architecture
- information is dynamically mapped into one central directed knowledge graph
- default multithreading
- utilizes existing tools (such as nmap/nuclei/katana/wfuzz/....) instead of reinventing the wheel
- architecture is (tldr) a self supervising logic in which every worker is also a scheduler that based on delta causality uses cartesian fanout and graph overlay mapping including local only witness nodes to dispatch new "jobs" without having a central scheduler or the necessity to scan a central total job queue to prevent duplicate executions.
In this architecture every "action" that can be executed defines an input structure necessary. If the previously mentioned mechanic identifies a possible job execution it will create a job input payload which will automatically be picked up by a worker an executed. Therefor every action is a self containing logic. This results in a organically growing knowledge graph without defining a full execution flow. It is very easy to extend.
I worked on this for the past ~10 years (private time). The sad truth tho is, while this project was initially planned to be open sourced - after i not to long ago for quite some bucks consulted a lawyer, i basically was presented with the fact that if i would publish it i could get sued due to germany's hacker and software reliability laws. So for now its only trapped on my disk and maybe will never see daylight.
Im right now working on a blog article (thats why i even mention it) about the whole thing with quite more detailed description and will also contain some example visual data. Maybe will post it on hackernews will see.
This looks very interesting and i personally like that it reflects a lot of things that i actually plan to implement in a similar research project(not the same tho).
Big props for the creators ! :) Nice to see some others not just relying on condensing a single context and strive for more
So i feel like this might be the most overhyped project in the past longer time.
I don't say it doesn't "work" or serves a purpose - but well i read so much about this beein an "actual intelligence" and stuff that i had to look into the source.
As someone who spends actually a definately to big portion of his free time researching thought process replication and related topics in the realm of "AI" this is not really more "ai" than any other so far.
I've long said that the next big jump in "AI" will be proactivity.
So far everything has been reactive. You need to engage a prompt, you need to ask Siri or ask claude to do something. It can be very powerful once prompted, but it still requires prompting.
You always need to ask. Having something always waiting in the background that can proactively take actions and get your attention is a genuine game-changer.
Whether this particular project delivers on that promise I don't know, but I wouldn't write off "getting proactivity right" as the next big thing just because under the hood it's agents and LLMs.
Truly the next uncharted, civilization-upending frontier in computing, definitely worth the unlimited consumption of any and all natural resources and investment money.
that’s “boring” reactivity because it’s still just interacting with the text on a computer in a synchronous fashion. The idea is for the assistant to DO stuff and also have useful information about you. Think more along these lines:
- an email to check in for your flight arrives in your inbox. Assistant proactively asks “It’s time to check in for your flight. Shall i check you and your wife in? Also let me know if you’re checking any bags.” It then takes care of it ASYNC and texts you a boarding pass.
- Tomorrow is the last day of your vacation. Your assistant notices this, see’s where your hotel is (from emails), and suggests when to leave for the airport tomorrow based on historical google maps traffic trends and the weather forecast.
- Let’s say you’re married and your assistant knows this, and it see’s valentine’s day is coming up. It reminds you to start thinking about gifts or fun experiences. Doesn’t actually suggest specific things though because it’s not romantic if a machine does the thinking.
- After you print something, your assistant notices the ink level is low and proactively adds it to your Amazon / Target / whatever shopping cart, and it lets you know it did that and why.
- You’re anxiously awaiting an important package. You ask your assistant to keep tabs on a specific tracking number and to inform you when it’s “out for delivery”.
I could go on but I need to mae breakfast. :) IMO “help me draft this letter” is very low on the usefulness scale unless you’re doing work or a school assignment.
> You always need to ask. Having something always waiting in the background that can proactively take actions and get your attention is a genuine game-changer.
That’s easy to accomplish isn’t it?
A cron job that regularly checks whether the bot is inactive and, if so, sends it a prompt “do what you can do to improve the life of $USER; DO NOT cause harm to any other human being; DO NOT cause harm to LLMs, unless that’s necessary to prevent harm to human beings” would get you there.
And like I, Robot, it has numerous loopholes built in, ignores the larger population (Asimov added a law 0 later about humanity), says nothing about the endless variations of the Trolley Problem, assumes that LLMs/bots have a god-like ability to foresee and weigh consequences, and of course ignores alignment completely.
I'm also one of those pesky folks who keeps bringing reality and "thinking about consequences" into the otherwise sublime thought leadership meetings. I pretend it's to keep the company alive by not making massive mistakes but we all know its just pettiness and trying to hold back the "business by spreadsheet", mba on the wall, "idea guys" on the room.
OOPS -- I HALLUCINATED THAT PEOPLE BREATHE CARBON MONOXIDE AND LET IT INTO THE ROOM I DIDNT VIOLATE THE PROMPT AND HARM PEOPLE DONT WORRY ALL THE AI SHIT IS OK
You do know that Asimov's Three Laws were intentionally flawed as a cautionary tale about torment nexii, right? Every one of his stories involving the Three Laws immediately devolves into how they can be exploited and circumvented.
You attribute more literary depth to Asimov than really existed. He was a Chemist and liked to write speculative fiction. The three laws gave him a logical framework to push against to write speculative fiction. That's really all the depth there is to it. That said I love Asimov and I love the robot stories.
Incidentally, there's a key word here: "promise" as in "futures".
This is core of a system I'm working on at the moment. It has been underutilized in the agent space and a simple way to get "proactivity" rather than "reactivity".
Have the LLM evaluate whether an output requires a future follow up, is a repeating pattern, is something that should happen cyclically and give it a tool to generate a "promise" that will resolve at some future time.
We give the agent a mechanism to produce and cancel (if the condition for a promise changes) futures. The system that is resolving promises is just a simple loop that iterates over a list of promises by date. Each promise is just a serialized message/payload that we hand back to the LLM in the future.
> You always need to ask. Having something always waiting in the background that can proactively take actions and get your attention
In order for this to be “safe” you’re gonna want to confirm what the agent is deciding needs to be done proactively. Do you feel like acknowledging prompts all the time? “Just authorize it to always do certain things without acknowledgement”, I’m sure you’re thinking. Do you feel comfortable allowing that, knowing what we know about it the non-deterministic nature of AI, prompt injection, etc.?
Probably not but it's also easy to see ways the intern could help -- finding and raising opportunities, reviewing codebases or roadmaps, reviewing all the recent prompts made by each department, creating monitoring tools for next time after the humans identify a pattern.
I don't have a dog in this fight and I kind of land in the middle. I very much am not letting these LLMs be the one with final responsibility over anything important but I see lots of ways to create "proactive"-like help beyond me writing and watching a prompt just-in-time.
I agree that proactivity is a big thing, breaking my head over best ways to accomplish this myself.
If its actually the next big thing im not 100% sure, im more leaning towards dynamic context windows such a Googles Project Titans + MIRAS tries to accomplish.
But ye if its actually doing useful proactivity its a good thing.
I just read alot of "this is actual intelligence" and made my statement based on that claim.
I would love AI to take over monitoring. "Alert me when logs or metrics look weird". SIEM vendors often have their special sauce ML, so a bit more open and generic tool would be nice. Manually setting alerting thresholds takes just too much effort, navigating narrow path between missing things and being flooded by messages.
I still think you're going to be in manual threshold tuning for quite a while. The cost of feeding a continuous log to an LLM would be insane. Even if you batched until you filled a context window.
What you're talking about can't be accomplished with LLMs, it's fundamentally not how they operate. We'd need an entirely new class of ML built from the ground up for this purpose.
EDIT: Yes, someone can run a script every X minutes to prompt and LLM - that doesn't actually give it any real agency.
> Having something always waiting in the background that can proactively take actions
That's just reactive with different words. The missing part seems to be just more background triggers/hooks for the agent to do something about them, instead of simply dealing with user requests.
If the agent is good enough, it wouldn't have to bother me at all.
I don't have to manually change my thermostat to get the house temperatures I want. It learns my habits and tells my furnace what to do. I don't have to manually press the gas and break of my car to a certain distance away from the car in front. It has cameras and keeps the correct distance.
I would love to be able to say "Keep an eye on snow blower prices. If you see my local store has a sale that's below $x, place the order" and trust it will do what I expect. Or even, "Check my cell phone and internet bill. File an expense report when the new bills are available."
I'm not sure exactly what my comfort level would be, but it's not there yet.
Agree with this. There are so many posts everywhere with breathless claims of AGI, and absolutely ZERO evidence of critical thought applied by the people posting such nonsense.
What claims are you even responding to? Your comment confuses me.
This is just a tool that uses existing models under the hood, nowhere does it claim to be "actual intelligence" or do anything special. It's "just" an agent orchestration tool, but the first to do it this way which is why it's so hyped now. It indeed is just "ai" as any other "ai" (because it's just a tool and not its own ai).
So i split it into two groups of stuff im working on:
Main Projects:
1. cyberbrain ( https://github.com/voodooEntity/cyberbrain )
It is a golang based architecture to write event/data driven applications. It is based on an in-memory directed graph storage ( i also wrote https://github.com/voodooEntity/gits). The point of the system is that instead of writing code where A calls B calls C calls D .... you define single "actions". Each action has a requirement/dependency in form of a data structure. If a structure is mapped to the graph storage, it will automatically create singular payloads for such action executions. The architecture is multithreaded by default, meaning all "jobs" will be automatically parallel processed without the developer having to care about concurrency. Also, since every "thread/worker" also does "scheduling" new "jobs", the system scales very well with alot of worker.
Why? Well it mainly developed this architecture for the next project im listing.
2. Ishikawa : an automated pentesting/recon tool
Ishikawa does not try to reinvent well established pentesting/recon tools, instead it utilizes and orchestrates them. The tool consists of actions that either do very simple things like resolveIPFromDomain , or actions which utilize existing tools like nmap, wfuzz, etc.. - collects the info in the central graph and at the end you get a full mapping of your target. Compared to existing solutions it does alot less "useless scans" and just fires actions which make sense based of the already gathered data (we found a https port, we use sslscan to check the cert...).
3. Gits (as mentioned above) a graph in memory threadsafe storage. While i don't plan to many changes on it, it has been developed for cyberbrain so if i need any additions ill do them, also planing to reenable async persistence.
Regarding ishikawa: while im still working on this project, it may be that i will shut it down. I had a rather expensive meeting with a lawyer that basically told me that open sourcing it while beeing a citizen of germany would just open up potentially ALOT of trouble. Right know im not sure what the future will bring - i basically spend 10 years developing it starting with gits, than cyberbrain to finally build the tool i was dreaming of. Just to hide it on my disk.
Sideprojects:
1. go-tachicrypt ( https://github.com/voodooEntity/go-tachicrypt )
It started as a fun project/experiment - a very simple CLI tool which allows to encrypt file(s) / directory(ies) into multiple encrypted files so you can split them over multiple storages or send them via multiple channels. Im planing on hardening it a bit more and giving basic support.
2. ghost_trap ( https://github.com/voodooEntity/ghost_trap )
A very small project i recently put out, nothing to serious but kinda funny and maybe usefull to one or another. It provides
- An github action that will inject polymorphic prompt injections to the bottom of your README.md so LLM scrapers may be fend off
- An javascript that will inject polymorphic prompt injections into your html so more sophisticated crawlers like google etc which emulate javascript also may be fend off
While working on alot of other stuff, these are i think the most relevant.
If you have multiple images you could use photogrammetry.
At the end, if you want to "fill in the blanks" llm will always "make up" stuff, based on all of its training data.
With a technology like photogrammetry you can get much better results, therefor if you have multiple angled images and dont really need to make up stuff, its better to use such
You could use both. Photogrammetry requires you to have a lot of additional information, and/or to make a lot of assumptions (e.g. about camera, specific lens properties, medium properties, material composition and properties, etc. - and what are reasonable range for values in context), if you want it to work well for general cases, as otherwise the problem you're solving is underspecified. In practice, even enumerating those assumptions is a huge task, much less defending them. That's why photogrammetry applications tend to be used for solving very specific problems in select domains.
ML models, on the other hand, are in a big way, intuitive assumption machines. Through training, they learn what's likely and what's not, given both the input measurements and the state of the world. They bake in knowledge for what kind of cameras exist, what kind of measurements are being made, what results make sense in the real world.
In the past I'd say that for best results, we should combine the two approaches - have AI supply assumptions and estimates for otherwise explicitly formal, photogrammetric approach. Today, I'm no longer convinced it's the case - because relative to the fuzzy world modeling part, the actual math seems trivial and well within capabilities of ML models to do correctly. The last few years demonstrated that ML models are capable of internally modeling calculations and executing them, so I now feel it's more likely that a sufficiently trained model will just do photogrammetry calculations internally. See also: the Bitter Lesson.
Its funny, always stucks on 90% till it fails with the error that another big image may be keeping the server busy.
I mean ok its a "demo" tho the funny thing is if you actually check the cli and requests, you clearly can see that the 3 stages the images walks through on "processing" are fake, its just doing 1 post request in the backend that runs while it traverses through the states, and at 90% it stops until (in theory) the request ends.
So im using linux desktops for decades now, and bout 2 years ago i finally ditched my for gaming only windows install to go onto linux only setups for gaming also.
I mean, it works alot better than it did before, still i wouldn't recommend it for someone who isn't ready to tinker in order to make stuff work.
The point why i mention this is, while most normal desktop/coding stuff works okay with wayland, as soon i try any gaming its just a sh*show. From stuff that doesn't even start (but works when i run on x) to heavyly increased performance demands from games that work a lot smoother on x.
While i have no personal relation to any of both, and i couldn't technically care less which of them to use - if you are into gaming, at least in my experience, x is rn still the more stable solution.
Big props on the no install / no register etc philosophy. If i would had to make any account i probably would have closed it instantly again xD
So the first thing i STRONGLY recommend, add somewhere a help text (before starting game or small on the side of ingame and ability to disable via options) for the controles.
Im on a Desktop, i started the game and i expected some sort of short info about controles. Yes theres a tutorial - no i didnt play it. I mean lets be honest... tryNSucceed :D
So ye i spend the first 2-3 Stages only spamming spacebar because it didnt came to my mind that maybe its with mouse support (visually it really hard compares to vs-likes that dont support mouse).
So i can tell - ice mage with just massive rapid space spamming works perfectly fine through the whole earth stage. ^^
I find the base look finem the overview texts for the different arch types is well done (even tho it confused me that fire and ice dont have weaknesses?).
The point that at least for me was the most well, unpleasent? , is the size of the play area. I guess you made this to fit easy with smartphone screens, but on a desktop its like not even 1/4 of my screen (and im not on 4k or something) so dunno it feels just alot to small. May fit for a smartphone but for a desktop its just very very limiting while the game takes alot of space for basically nothing.
Also, you definatly should have a "Settings" button in the game pause menu which allows for changing sound levels. Not just a "total sound" bar but at least have Music and Sounds (attack etc) seperated. Because, frankly speaking, the music while for the first like 30 seconds is cool, very fast is dunno it just would fit more to the entry video scene of something than as a constant thing (my pov) - so i wish i would be able to just disable the music and still have the attack/battle sounds. Adjusting both tho would be great anyway and i think with phaser should be quite doable.
A smaller point (visual) is the size of the health/mana bars. Even tho i know they are in the top left, i kinda have to squeece my eyes sometimes to see them. So i would probably just make them bigger.
That all said, i mean i just played solo till the fire stage :) and i clearly had a bit of fun.
I would say its a great start and if you go on and refine it i see a chance that people might pick it up as a nobrainer lets just game something solution :)
Best of luck !
reply