david_d8912
david_d8912 on Hacker News
For curiosity: why Hermes? is there any part that you think it excels Claude or Codex (or it's just easier access from Telegram)
Interesting tool. A few suggestions with a quick glance:
1. can we make it representing daily real-time data by hook-up e.g. RSS, twitter? My personal use case is a "radar" would apply to more fresh info. The known tasks will be on a kanban (which also split priority or tag clearly)
2. not sure if it'll be better in-app, but quickly check the screenshot, the object titles are a bit small on radar view. It should be more readable.
GPT-5.5 + Opus-4.7 here. Codex for pure coding task with clear goal, claude code for the rest. Also combined with opencode to experiment new models.
fyi: I didn't have much lock on Deepseek v4 pro, with opencode + openrouter it's incredibly slow. How did op did it?
cool. I agree with all the front points. The last part ```Keep behavior instructions positive where possible```, do you have good experience on it. I'm only asking since my own experience is they're constantly not followed.
The project directory, sure. I'm more of talking about behavior rules here. What are you guys writing, is it effective
Posted by
david_d8912 on Saturday May 16, 2026 @10:33AM
from the ask-hn dept.
For people using coding agents regularly: do you still invest time in user-level and repo-level instruction files like CLAUDE.md, AGENTS.md, etc.?It's not a surprise that you don't expect the rules there to be followed even with less than 100 lines of changes. Yet still see folks like Karpathy post rules around with hundreds of stars.Tell me if you're still spending the effort of trying...
It's already been the case for Obsidian? you create a github repo, create notes as md files, obsidian plugin sync it?
Actually not obsidian, but any note viewer like app. You put files in Google Drive or git repo for that matter.
So it's a cli agent swarm? any difference to the omc, everything-cc swarm features
Yeah that's what I'm doing, upfront work & code review. But it drastically slowing down the dev speed.
Posted by
david_d8912 on Wednesday May 06, 2026 @04:37AM
from the self-post dept.
I'm using both claude code (opus-4.7) and codex (gpt-5.5). The agents are perfectly capable of delivering most features hands free these days, but the code quality is still miserable without another few rounds of prompt. And it's worse for personal vibe project without well-written history code files.Simple example would be a poor extension of existing util methods, agent always duplicate and...
maybe a consequence of saving GPU for newer models? Also tuning effort level suppose to help, haven't get enough dp on this though
Now it'll be interesting to see if the AI companies do the same