[HN Gopher] Kilo Code: Speedrunning open source coding AI
       ___________________________________________________________________
        
       Kilo Code: Speedrunning open source coding AI
        
       Author : ofou
       Score  : 63 points
       Date   : 2025-03-26 16:15 UTC (6 hours ago)
        
 (HTM) web link (blog.kilocode.ai)
 (TXT) w3m dump (blog.kilocode.ai)
        
       | suddenlybananas wrote:
       | >This success taught me an important lesson: an extremely fast-
       | moving community can achieve incredible things
       | 
       | what a trite observation
        
         | janpaul123 wrote:
         | And yet few do it!
        
           | achierius wrote:
           | Is that true? Plenty of fast-moving communities have achieved
           | amazing things, both in computing and outside of it.
        
             | janpaul123 wrote:
             | Still a small percentage! Let's get more of this happening.
        
       | ilrwbwrkhv wrote:
       | What is so problematic now about Silicon Valley is that the true
       | lesson of hackers have been completely lost.
       | 
       | The primary purpose of the hacker mindset was protection against
       | groupthink and cargo culting. And now it seems all people in tech
       | only cargo cult and only groupthink.
        
         | hengheng wrote:
         | That ship sailed the moment the super-individualist hacker
         | turned into the 10x programmer.
        
           | NoOn3 wrote:
           | True super-individualist hacker won't use AI tools. :-)
        
       | heymax054 wrote:
       | $15 worth of Claude 3.7 tokens? Does this translate to 3 "hello
       | world" files?
        
       | bluelightning2k wrote:
       | I usually support everything but isn't this literally just "we
       | are trying to fork roo code and pay $15 of your tokens so we can
       | show VCs that we have users" - as in people like free money. But
       | that wouldn't be enough of a bribe to justify using the fork over
       | the real project for me at least
        
         | janpaul123 wrote:
         | Our backers have no interest in fake metrics. ;) It's a good
         | way to quickly get feedback, which is key to our strategy.
         | Totally fine to keep using Roo Code (or Cline) of course!
        
       | adamgordonbell wrote:
       | I talked to JP about this project. He's excited in a way that's
       | hard not to catch. His core thesis is simple: coding agents are
       | the future, and the winners will be the ones who can execute.
       | 
       | It won't be OpenAI or Claude. They have other priorities. The
       | real opportunity is for small teams who move fast, stay close to
       | users, and keep ahead of the pack.
       | 
       | That makes sense. LLMs are already powerful, almost magical at
       | times. But using them as coding agents still takes real work.
       | They can do amazing things, and be frustrating and make a mess.
       | There are rough edges and big gaps.
       | 
       | Those will get fixed. The question is who gets there first.
       | 
       | The counterpoint to that would be that all these tools are gonna
       | end up sort of the same and there won't be a way to
       | differentiate.
       | 
       | Which way will it play out? I'm not really sure.
        
         | janpaul123 wrote:
         | You're too kind!! The speedrun ethos has already been super fun
         | with this team. :)
         | 
         | I hope that we'll also be able to bring enough skills,
         | strategy, and taste to the space. Time will tell, but we're
         | giving it our best shot!
        
       | xnx wrote:
       | Gemini 2.5 seems to be the current king of AI coding. In addition
       | to being "smart", it has a huge context window. The one-shot
       | examples on Twitter are astounding.
        
         | outside2344 wrote:
         | Really? I would have said it was Claude-3.7 based on
         | experience.
        
           | Workaccount2 wrote:
           | Claude was until yesterday
        
       | janpaul123 wrote:
       | JP here! Would love to answer your questions!
       | 
       | We listed a bunch of ideas for larger improvements in the blog:
       | Instant app; Up-to-date docs; Prompt/product-first workflows;
       | Browser IDE; Local/on-prem models; Live collaboration; Parallel-
       | agents; Code variants; Shared context; Open source sharing; MCP
       | marketplace; Integrated CI; Monitoring/production agents;
       | Security agents; Sketching..
       | 
       | What would you like us to build?
        
         | arevno wrote:
         | The obvious thing would be LSP interrogation, which would allow
         | the token context to be significantly smaller than entire
         | files. If you have one file open, and you are working on a
         | function that calls out to N other modules, instead of packing
         | the context with N files, you get ONLY the sections of those
         | files the LSP tells you to look at.
        
           | janpaul123 wrote:
           | Yes! This is high on our list. Context window compression is
           | a big deal, and this is one of the main ways to do it, IMO.
           | 
           | Have you tried any tools that do this particularly well?
        
         | amarant wrote:
         | One thing that I think would be cool, and that could perhaps be
         | good starting point, is a TDD agent. How I imagine this
         | working:
         | 
         | User (who is a developer) writes tests, and a description of
         | the desired application. The agent attempts to build the
         | application, compiles the code, runs the tests, and
         | automatically feeds any compiler errors and test failures back
         | to agent so that it can fix it's own mistakes without input of
         | the user.
         | 
         | Based on my experience of current programming agents, I imagine
         | it'll take the agent a couple of attempts to get an application
         | that compiles and passes all the tests. What would be really
         | great to see is an agent (with a companion application
         | probably) that automates all those retries in a good way.
         | 
         | i imagine the hardest parts will be to interpret compiler
         | output, and (this is where things get real tricky) test output,
         | and how to translate that into code changes in the existing
         | code base.
        
           | janpaul123 wrote:
           | Yeah, this is a great workflow! What's more, agents are
           | particularly good at writing tests, since they're simpler and
           | mostly linear, so they can even help with that part.
           | 
           | As to your point of automating retries, with my last
           | prototype I played a lot with having agents do multiple
           | parallel implementations, and then pick the first one that
           | works, or lets you choose (or even have another agent
           | choose).
           | 
           | Have you tried any tools that have this workflow down, or at
           | least approach it?
        
         | spankalee wrote:
         | Your timeline is indeed crazy fast. Did you recruit the 9
         | others in your first week? Did you pitch and secure funding in
         | that week too? reply
        
           | janpaul123 wrote:
           | In roughly the last 2 weeks, yes. It helped that everyone
           | involved also activated their network, so we got a
           | multiplicative effect. Can't speak to funding for now
           | unfortunately.
        
         | 999900000999 wrote:
         | This is going to sound a bit odd, but I suggest you detail what
         | your tools do well and what they struggle with. For example I
         | love Haxe, which is a niche programming language primarily for
         | game development.
         | 
         | The vast majority of the time I try to use an llm with it, the
         | code is essentially useless as it will try to invent methods
         | that don't even exist.
         | 
         | For example if you're coding agents are really only good at
         | JavaScript and a little bit of python, tell me that front and
         | center.
        
           | janpaul123 wrote:
           | Good point! In that sense we're similar to most AI coding
           | agents in that the languages we do well are the languages the
           | mainstream LLMs do well. We might zoom in and add really good
           | support for particular languages though (not decided yet), in
           | which case we'll def mention that front and center!
           | 
           | Have you found any LLMs or coding agents that work well with
           | Haxe? It might be a bit too niche for us (again, not sure
           | yet), but I'd be very curious to see what they do well!
        
             | Zondartul wrote:
             | In my unqualified opinion, LLMs would do better at niche
             | languages or even specific versions of mainstream
             | languages, as well as niche frameworks, if they were better
             | at consultig the documentation for the language or
             | framework, for example, the user could give the LLM a link
             | to the docs or an offline copy, and the LLM would
             | prioritise the docs over the pretrained code. Currently
             | this is not feasible because 1. limited context is shared
             | with the actual code, 2. RAG is one-way injection i to the
             | LLM, the LLM usually wouldn't "ask for a specific docs
             | page" even if they probably should.
        
               | janpaul123 wrote:
               | 100% agreed on both points. Point 1 relates to
               | https://news.ycombinator.com/item?id=43486526 as well.
               | It's one of the biggest challenges, though maybe it'll
               | automatically get better through models with bigger
               | context windows (we can't assume that though)?
        
             | 999900000999 wrote:
             | https://www.greptile.com/
             | 
             | This works well, however it literally will need to digest
             | an entire repository. So for example if I feed it a
             | repository for a haxe framework, it'll work much better
             | than something like Chat GPT.
        
               | janpaul123 wrote:
               | Thanks! That does look like a great tool.
        
       | cpldcpu wrote:
       | Their approach seems very compelling, but I don't understand
       | if/how they are building a differentiated product? The space of
       | code agents is already pretty crowded.
        
         | janpaul123 wrote:
         | We'll take all the features people love in other products, and
         | implement them in a coherent package as quickly as we can.
        
       | handfuloflight wrote:
       | "Our goal is to rapidly make the software better, not to have a
       | shiny website."
       | 
       | Weird flex, but OK.
        
         | rpmisms wrote:
         | Excellent flex. The purpose of a thing is what it does.
        
           | handfuloflight wrote:
           | More context, the statement is in reply to the question they
           | posed on their own site: "Why is this website so ugly?"
           | 
           | First, I don't think the website is ugly per se. Second, the
           | weird flex is assuming that a website which had more effort
           | put into it than what they put into theirs is "a shiny
           | website."
           | 
           | Design aside, _there 's absolutely no statements regarding
           | what makes this product differentiated._ So it doesn't even
           | succeed on its own terms.
        
       | tristor wrote:
       | To be honest, I have yet to use any GenAI tool that makes me feel
       | like it can replace me just writing code (I write this as an
       | Engineer turned PM, that would really like the promise of GenAI
       | to be true). What I'd actually like to see more than anything is
       | a GenAI "agent" that can act like the /user/ of my software to
       | help me identify gaps in documentation as the software changes
       | and the documentation drifts/becomes stale, and generally help me
       | to explore code paths that are off the happy path but will get
       | hit by real users. I think there's a lot more value in having
       | GenAI help me test/document my work than in trying to do my work,
       | because I will always write higher quality code than GenAI can
       | produce.
        
         | janpaul123 wrote:
         | Totally agree!
        
       | rounce wrote:
       | Is this a real open source project or a pretend 'source (maybe-
       | kinda) available' kinda thing where the really useful part is
       | stuffed behind a paywall and the 'open source' part is just to
       | lure you into the walled garden?
        
       | realharo wrote:
       | _> We want to build for the dream of billions of programmers;
       | billions of artists; billions of scientists--using computing as
       | moldable clay._
       | 
       | At that point, why even keep humans in the loop. Just let it
       | exist in the background and generate better ideas than any human
       | would anyway.
        
         | observationist wrote:
         | Just let the car go wherever it wants, faster than any human
         | would anyway. Or - just let the fire exist in the background,
         | it'll generate more heat than humans will ever need anyway.
         | 
         | The point isn't to make humans pointless. The point is to
         | empower humans. We need to remain in control, and be the users
         | of the tool, and not a tool for some mindless system.
         | 
         | Intelligence and consciousness are separate things - you can
         | automate a lot of intelligence without having even rudimentary
         | consciousness or self awareness - LLMs currently in operation
         | are at most pseudo-conscious within their test-time contexts,
         | and even then, every pass resets whatever awareness there might
         | be. With millions of tokens context length, that might start to
         | enter into the realm of a thing we should be concerned about,
         | but even then, there's no ongoing persisted state to carry
         | anything between passes aside from the text or image patch
         | tokens or what have you.
         | 
         | What this means, essentially, is that we can augment our human
         | capabilities without usurping the agency of some artificial
         | being - these AIs are not individual moral agents in their own
         | right, and likely will never be unless we specifically build
         | that recursion and persistent state into the models, and
         | incorporate a realtime adaptive self and world construct.
         | 
         | This means that the software is a tool - use the tool to
         | augment your life and be a force multiplier in everything you
         | do. The scope of intelligence augmentation has leapt from
         | spreadsheets to nearly every cognitive domain in the human
         | experience - people proficient with excel were better
         | accountants than people using pen and paper. People using
         | delivery vans are better than people using a horse and wagon.
         | This new technology means that people using AI will be able to
         | do more, faster, and likely better, than people who don't.
         | 
         | With neural lace - whatever form it ends up being - we'll end
         | up with genuine exocortex augmentation. Even without that
         | direct integration, however, the human in the loop is the
         | entire point of this technology. There's a tiny list of things
         | conscious machines might be good for, and all sorts of deep and
         | obvious arguments for not creating a new, self aware, agentic
         | species that's immediately in conflict with and on a trajectory
         | to outcompete humans.
         | 
         | Use the tool of AI to be a force multiplier for everything in
         | your life that AI is capable of handling well. This makes you a
         | benevolent dictator for life in your own life, delegating
         | everything that makes sense, working with it to free up your
         | resources for the things that you decide are the highest
         | priority. Spend more time brainstorming, building
         | relationships, deploying resources, and getting the most out of
         | being human. This is the promise of AI, and why people get
         | excited about it. We're going to have a huge struggle, as
         | humanity, in dealing the empowerment and amplification of
         | everything in our lives. Making sure that we retain agency,
         | that humans are ultimately in charge of our own destiny, is
         | probably the most important principle to adhere to, above all
         | others.
        
           | realharo wrote:
           | I was mainly referring to phrases such as _" billions of
           | scientists"_ - the point of science is to solve problems and
           | discover knowledge. If you have an AI good enough to achieve
           | that (billions of scientists), that means it can probably
           | progress without being actively driven by people at all - and
           | probably do a better job at it too.
           | 
           | We can still do things "for fun", but our efforts will be
           | more toys than serious projects (except when it comes to
           | relationships with other people).
        
       | quikoa wrote:
       | >We don't take any cut, either per token or per top-up. In the
       | future we'll add more LLM providers.
       | 
       | So where does the money come from?
        
         | janpaul123 wrote:
         | At this point we plan to monetize enterprise features (LDAP
         | login, things like that).
        
       ___________________________________________________________________
       (page generated 2025-03-26 23:00 UTC)