ac-ciano

4 ms·

1.
by ac-ciano 10d ago
I replied to everyone on my own.
2.
by ac-ciano 10d ago
I agree with you, I did not reply to anybody with an LLM. I employed AI for building
3.
by ac-ciano 10d ago
Sure, as you could on OpenCode/PI, but most of us use subscriptions for which you are forced to use the relative UI
4.
by ac-ciano 11d ago
indeed, but I also value the freedom of using whichever provider I want and not get locked into Claude Code alone for instance. I want to use both Opus, DeepSeek and GPT 6 at the same time, not just the 2/3 models Anthropic offers. Wha
5.
by ac-ciano 11d ago
very interesting, the concept is the same. I see they flagged Opus as the suggested planner, which makes a lot of sense. Thanks
6.
by ac-ciano 11d ago
nope but I will check it out, thanks
7.
by ac-ciano 11d ago
Absolutelly, it is brilliant! I use it daily myself from Ordewell
8.
by ac-ciano 11d ago
Yeah absolutelly. I just wanted that exact framework in a simple controllable UI, where I could also plug in any of my subscriptions/API keys such as Claude Code or Codex at the same time. With Ordewell you get all layed out as tasks t
9.
by ac-ciano 11d ago
I like your analogy. The main problem though is context and keeping it clean as much as possible as long a parallelization. This is what drove to build this tool: having control of everything that the LLMs will do, controlling all with one
10.
by ac-ciano 11d ago
I thought about it but where the tool shines are large undefined tasks, which are complex to quantify and test (not as easy as implementing a simple bug fix that you can test directly). Even if I were to find such test dataset it would prob
11.
by ac-ciano 11d ago
Yeah it's exactly what I've seen as well. Ordewell is close to your high end case: a frontier model builds the plan once, then it's fixed, the executing model can't reinterpret it. Won't get you to 4B, but should h
12.
by ac-ciano 11d ago
To me that's not the embarrassing part... unmaintainable would be. Using AI for coding and writing doc is standard practice now, not using it would be insane.
13.
by ac-ciano 11d ago
Fair, and I won't pretend otherwise. I heavily used AI to help write the docs and code. The thing is though... I designed it and I stand behind it.
14.
by ac-ciano 11d ago
Fair — AGENTS.md is prose the model has to re-interpret every session, and that reinterpretation is exactly where weaker models lose the thread. Here the plan is parsed and enforced as structured data: tasks with declared dependencies and o
15.
Show HN: Ordewell – turn one goal into an ordered plan of coding-agent tasks(github.com)
50 points by ac-ciano 11d ago|30 comments
16.
by ac-ciano 2mo ago
At the end of the day, OpenAI built the system and provided the instructions that allowed this gap to exist. Relying on instruction-following rather than strict, non-negotiable code boundaries for tool use is always going to lead to securit
17.
by ac-ciano 2mo ago
Pure statistics and AST history over throwing another LLM at the problem is such a refreshing approach. Super clever project, congrats!