TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Show HN: Atari Missile Command Game Built Using AI Gemini 2.5 Pro

136 点作者 vbtechguy大约 1 个月前
A modern HTML5 canvas remake of the classic Atari game from 1980. Defend your cities and missile bases from incoming enemy attacks using your missile launchers. Initially built using the Google Gemini 2.5 Pro AI LLM model.

25 条评论

greybox大约 1 个月前
One question I continue to ask myself when I see impressive AI examples like this, (because it is impressive) is &quot;are we solving the right problems&quot;? Instead of training an AI to produce a tonne of JS to create something that&#x27;s been done before, should we not be engineering better tools to allow people to precisely craft software without the need for AI.<p>I worry that because we are now able to instantly produce a bunch of JS to do X thing, we will be incentivized not to change the underlying tools (because one, only AI&#x27;s are using it, and two AI&#x27;s won&#x27;t know how to use the new thing)<p>I worry this will stall progress.
评论 #43652202 未加载
评论 #43652334 未加载
评论 #43666138 未加载
评论 #43658929 未加载
eknkc大约 1 个月前
This is impressive honestly.<p>I also find it interesting that the ai code submissions like this one are generally vague about the process.<p>This seem to be created using Cline on VsCode, prompting to Gemini 2.5 Pro using OpenRouter.<p>The commit history implies that a crude version was created by the LLM using an initial prompt and then gradually improved with features, fixes etc. Assuming ongoing conversations with the AI agent.<p>All code in a single index.html file which might not be great for human coders but who cares to be fair.<p>All in all, a prompt history would be really educational if anyone thinks about doing something similar.
评论 #43651543 未加载
评论 #43651664 未加载
评论 #43652658 未加载
评论 #43651610 未加载
评论 #43653994 未加载
评论 #43651999 未加载
golden-gato大约 1 个月前
I am getting the following error when running the AI analysis: AI analysis received, but failed to parse structured advice. Raw content: &quot;```json {&quot;summary&quot;: &quot;The player, LIGMA, demonstrates a high rate of missile usage, firing all 170 missiles by wave 4. Despite destroying a significant number of enemies, including two shield bombs and 23 planes, LIGMA lost a substantial amount of cities and bases, indicating potential issues with defense coverage or target prioritization.&quot;, &quot;advice&quot;: [&quot;Prioritize city defense. Focus on intercepting missiles and bombs headed directly for the cities, as losing four cities by wave 4 significantly hinders the score potential.&quot;, &quot;Increase accuracy for bonus points. With only 11 accuracy bonus hits out of 170 missiles fired, practice aiming closer to the incoming targets to maximize the accuracy bonus and the score multiplier.&quot;, &quot;Consider purchasing and using special weapons like the mega bomb or sonic wave, as the click data doesn&#x27;t reflect any usage of these options which can become invaluable in later waves.&quot;, &quot;Improve resource management by not firing more missiles than is needed to destroy any given incoming attacker. The player fired all 170 missiles indicating a poor kill&#x2F;shots ratio.&quot;], &quot;improve target evaluation speed and accuracy.&quot;} ```&quot;<p>Could not parse specific advice due to formatting issues in the AI response content.
评论 #43652430 未加载
franze大约 1 个月前
If this is too stressful, I created my own star based relaxation game yesterday before boarding a plane. Sadly did not get much traction. Its basically the travelling salesman in space<p><a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=43648528">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=43648528</a>
评论 #43651908 未加载
评论 #43658585 未加载
spacecadet大约 1 个月前
I have been building a &quot;realistic&quot; but simplified 2D universe simulator since GPT3.5 was first introduced, it has slowly grown to become a 1000loc Python file that uses Pygame. I think it represents a good mix of what might appear in training as code examples and physics understanding. Aside from occasionally &quot;playing&quot; the game, its a zero-player game... I mainly use it to test new models. Can the model handle 1000 lines of complex physics code? Can the model make improvements to areas that I already know are poorly optimized or incorrect?
评论 #43654195 未加载
jjmarr大约 1 个月前
Congrats to &quot;Kitten&quot; for being the first to cheat the high score.
Apfel大约 1 个月前
Sorry - couldn&#x27;t find a security.txt but thought I should alert you your API is currently vulnerable to prompt injection on at least a few of the fields in the structured JSON you send when analysing the game.<p>Happy to give more details if there&#x27;s any way to get in touch outside this thread.
评论 #43660379 未加载
eesmith大约 1 个月前
I got up to level 8 before I decided to go to the store.<p>I had something like $500,000. I bought up the entire inventory (at least, until the Buy buttons stopped working - there were still items available).<p>It then became a &#x27;click as fast as you can and don&#x27;t care about strategy&#x27; game, so I stopped.
评论 #43651590 未加载
评论 #43609519 未加载
indigodaddy大约 1 个月前
Any chance you could reveal your entire initial prompt, or even your later-down-the-line detailed portable prompt that you mentioned you could utilize from one LLM chat to another?
exabrial大约 1 个月前
Had a great conversation today with Gemini today where it told me 90000us is 9ms
Oras大约 1 个月前
Using `Tab` out of all the keyboard keys is such a weird option!<p>What was the idea behind this choice?
nottorp大约 1 个月前
Curious... if anyone can read the source fast...<p>Are the game state updates and input coupled with the frame rate, as in most beginner game development tutorials out there, or did the &quot;AI&quot; do the right thing and decouple them?
评论 #43664090 未加载
kenjackson大约 1 个月前
What prompt did you use and how much did you have to code on your own?
评论 #43651029 未加载
评论 #43651551 未加载
评论 #43651006 未加载
jansan大约 1 个月前
&gt; Initially built using the Google Gemini 2.5 Pro AI LLM model.<p>What does this mean?
评论 #43664809 未加载
评论 #43651122 未加载
m3kw9大约 1 个月前
Is a lossy compressed game extracted with lossy error correction
Tistron大约 1 个月前
After some levels I could just press sonic wave 3 times and win without even doing the clicking. Had enough money to replenish. After winning level 29 nothing more happened.
echoangle大约 1 个月前
Does missile command not have friendly fire? Can you really detonate missiles directly over your own bases? Currently, it seems pretty easy.
评论 #43652054 未加载
kookamamie大约 1 个月前
The background is animating far too fast and is very noisy to look at. Win&#x2F;Chrome&#x2F;9950x3D&#x2F;RTX5090.
评论 #43668111 未加载
indigodaddy大约 1 个月前
How&#x2F;why are random sites&#x2F;apps hosted on the centminmod.com domain?
评论 #43664795 未加载
jrvarela56大约 1 个月前
That was fun, thank you!
DocTomoe大约 1 个月前
When you get old enough, you realise the cycle turns around and around<p>A new dev tech shows up. Old devs say: That&#x27;s not <i>real</i> programming. Real programmers use the old ways. You take all the skill out of it. You&#x27;ll never learn to do it the &quot;right way&quot;. And then, it becomes standard, an no-one wants to go back but a few hobbyists.<p>It&#x27;s been this way with switching from assembly to more human-readable languages.<p>It&#x27;s been this way with syntax highlighting.<p>... and with IDEs.<p>I remember when we scoffed at IntelliSense over the Water Cooler because them kids didn&#x27;t have to memorise stuff anymore.<p>I kept cursing at Docker and npm insanity, having colourful languages for people who hid behind abstraction because they did not understand basic functionality.<p>And today, it is AI. Right now, it divides. Those who love it, those who consider it &#x27;cheating&#x27; or &#x27;stealing other people&#x27;s code&#x27;. In reality, it is just another level of abstraction in the development stack. Tomorrow, it&#x27;ll just be &#x27;the standard way to do things&#x27;.<p>I wonder what comes next.
vrighter大约 1 个月前
does it also run in less than 128 bytes of ram, with half a scanline as a &quot;framebuffer&quot;?
mentalgear大约 1 个月前
I don&#x27;t know how impressive this is, since the exact or very close variations of these games certainly can be found in the model&#x27;s training data.<p>This is neither innovative nor showing creativity IMO and reminds me more of twitter hype-bro posts, than something truly HN front-page worthy.
评论 #43654384 未加载
评论 #43652147 未加载
评论 #43652326 未加载
评论 #43651615 未加载
评论 #43652492 未加载
评论 #43652070 未加载
irjustin大约 1 个月前
I started playing with Gemini 2.0 Pro and 2.5 is another level. It&#x27;s my goto for coding over Cursor. I just wish they put more effort into the VSCode plugin&#x2F;agent.
评论 #43651983 未加载
dailykoder大约 1 个月前
I started playing with vim, but neovim is another level. It&#x27;s my goto for coding over vim. I don&#x27;t expect them to put more effort into anything, because it just works as a editor&#x2F;IDE.
评论 #43651376 未加载