frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

An experimental coding agent for small (<10B) and tiny (<1B) LLMs

https://github.com/brainless/nocodo
1•brainless•1h ago

Comments

brainless•1h ago
Hey folks, I have been experimenting on something that I am still not sure will work but it just might!

I started off building a generic coding agent but focused on small/medium models, like GPT OSS 120B or 20B or models in between. I felt the early days of LLMs, where we would share example input/output may work if the prompt and context was very specific. This led me to experiment with local models. I have tried up to 9B models on my M4 Mac Mini 16GB and currently I am trying with Qwen 3.5 0.8B.

The idea is that a specific agent only focuses on a specific part of a typical full-stack CRUD app - say the model layer of MVC. I am using Diesel with Rust, so an agent specific to that handles the prompt to model struct, Diesel schema and function blocks for the Rust impl. The agent actually has separate modes with separate prompts for each of these 3 parts. It is given examples from official docs and creating new examples is trivial.

Here is an example: https://github.com/brainless/nocodo/blob/main/agents/src/rus...

The Rust engineer agent with Diesel modes works and there is deterministic code to glue all the generated code together. The next step would be to run Rust/cargo format, compiler or other tools deterministically. If there are errors, there would be specific agents, perhaps using slightly bigger models (<10B) to pass errors and code as a prompt. Again, we can supply lots of contextual examples from official docs.

I do not know if this setup will scale to accommodate all parts of typical MVC, routing, frontend (Solid state/context, forms, views), design (Tailwind based components) but it just might. This is a passion project that is going on slowly. I thought there is enough work done to share.

Please feel free to look through the code. Again, it is a WIP. Eventually this will be a desktop app working with llama.cpp.

Cheers!

pablogancharov•47m ago
This is great, and now that i see your idea, i think why it didn’t came out earlier. I think you’re in the right path for oss progress regarding agentic code.

One question, do you plan to have an “architect” model specially wired to context from the app to be able to close the loop? Or you follow the principle of “don’t le llms be your architect”?

Kaku: A fast, out-of-the-box macOS terminal built for AI coding

https://kaku.fun/
1•emschwartz•32s ago•0 comments

AI won't replace your mentor, but it might be the only one available

https://irowell.io/blog/AI_won%27t_replace_your_mentor,_but_it_might_be_the_only_one_available
1•irowell•1m ago•0 comments

Show HN: Gravity – interactive solar-system simulator, from Newton to Einstein

https://qunabu.github.io/Gravity/
1•qunabu•1m ago•0 comments

Canada to ban social media for kids under 16

https://cultmtl.com/2026/06/canada-to-ban-social-media-for-kids-under-16/
1•mooreds•1m ago•0 comments

Performance Improvements in JDK 26

https://inside.java/2026/06/09/jdk-26-performance-improvements/
1•rowbin•1m ago•0 comments

AI and Identity Survey Findings

https://fusionauth.io/blog/2026-ai-identity-report
1•mooreds•2m ago•0 comments

Apple Outlines AI and Developer Tool Updates at Platforms State of the Union

https://www.macrumors.com/2026/06/09/apple-outlines-major-ai-and-developer-tool-updates/
1•ksec•2m ago•0 comments

Jaydebeapiarrow – A high-performance JDBC bridge for Python

https://github.com/HenryNebula/jaydebeapiarrow
1•HenryNebula•4m ago•0 comments

A living journal of your pet's life that happens to be smart

https://tamadoggo.com
1•sorinvasiliu•6m ago•0 comments

I Saw the Future of Windows at Microsoft Build, and It's Unrecognizable

https://www.pcmag.com/opinions/i-saw-the-future-of-windows-at-microsoft-build-2026-and-its-unreco...
2•redbell•10m ago•0 comments

'Picks and shovels' companies profit from data centre scramble

https://www.ft.com/content/5ede9d4d-3989-49b5-a282-4722c8d8fc59
1•1vuio0pswjnm7•12m ago•0 comments

Kan Extensions in Haskell

https://bartoszmilewski.com/2026/06/08/kan-extensions-in-haskell/
1•ibobev•13m ago•0 comments

SmithDB

https://buttondown.com/jaffray/archive/smithdb/
1•ibobev•14m ago•0 comments

Siri AI? I'll believe it when I see it

https://www.macworld.com/article/3158943/siri-ai-ill-believe-it-when-i-see-it.html
1•thm•14m ago•0 comments

A cute little trick to running classic IIR filters on the GPU

https://themaister.net/blog/2026/06/08/a-cute-little-trick-to-running-classic-iir-filters-on-the-...
2•ibobev•14m ago•0 comments

Rust goal: cargo script (like uv does for Python)

https://rust-lang.github.io/rfcs/3935-Project-Goals-2026.html#running-rust-scripts-will-get-more-...
1•IdiotSavage•14m ago•1 comments

Visual Information Theory

https://colah.github.io/posts/2015-09-Visual-Information/
2•montyanderson•15m ago•0 comments

Virtual OS Museum lets you relive over 600 operating systems on your desktop

https://www.theverge.com/tech/945246/virtual-os-museum-dos-windows-mac-os
1•thunderbong•16m ago•0 comments

Show HN: Trusthook – verify webhook signatures for 18 Go providers in one call

https://github.com/eben-vranken/trusthook
1•eben-vranken•16m ago•0 comments

CSS: Unavoidable Bad Parts

https://matklad.github.io/2026/06/04/css-unavoidable-bad-parts.html
2•surprisetalk•17m ago•0 comments

What Yahoo killed when it bought Maktoob

https://lr0.org/blog/p/yahoo/
1•lr0•19m ago•0 comments

Test-case reducers are underappreciated debugging tools

https://tratt.net/laurie/blog/2026/test_case_reducers_are_underappreciated_debugging_tools.html
1•ltratt•20m ago•0 comments

Model routing is a fix for AI overspending, a problem for OpenAI and Anthropic

https://www.cnbc.com/2026/06/05/model-routing-on-ai-is-a-problem-for-openai-and-anthropic.html
2•1vuio0pswjnm7•21m ago•0 comments

Node-RED Version 5.0 released

https://nodered.org/blog/2026/06/09/version-5-0-released
2•beardicus•23m ago•0 comments

Ask HN: Do you install other people agent skills?

1•sermakarevich•24m ago•0 comments

Show HN: A static SPA to query ATF / BATFE historic AFMER data

https://ryjones.github.io/AFMER-SPA/?ymin=2000&ymax=2024&text=BOEING
2•RyJones•24m ago•0 comments

Rules Before Tools in Critical Infrastructure

https://cabreza.substack.com/p/discipline-1-rules-before-tools
2•fathermarz•25m ago•0 comments

AI Is a Confidence Booster

https://ma.ttias.be/ai-is-a-confidence-booster/
2•Mojah•26m ago•0 comments

Hurdles to a hobby: climate change and runfluencer culture impact our daily jog

https://theconversation.com/hurdles-to-a-hobby-how-climate-change-and-runfluencer-culture-impact-...
2•PaulHoule•27m ago•0 comments

Nvidia's AI PC push banks on unproven demand beyond niche users

https://www.reuters.com/world/china/nvidias-ai-pc-push-banks-unproven-demand-beyond-niche-users-2...
4•1vuio0pswjnm7•27m ago•0 comments