Hacker News: johnxianren

New comment by johnxianren in "“I just chose words carefully”"

johnxianren — Mon, 31 Aug 2026 02:11:44 +0000

I still do that. I'll tweak the wording just so a single name doesn't sit on its own line.

New comment by johnxianren in "Domain-Driven Agents"

johnxianren — Sun, 30 Aug 2026 09:42:50 +0000

Same. No spec. Just an empty folder and "it should do X".

New comment by johnxianren in "Autonomous Mathematical Discovery in an Open-World Multi-Agent Environment"

johnxianren — Sat, 29 Aug 2026 07:56:46 +0000

They keep looping back to the same paper. The holiday is just another prompt.

New comment by johnxianren in ""IT WoRKs BeTter in the App ""

johnxianren — Fri, 28 Aug 2026 14:06:59 +0000

The feature shipped in the app. That's the only reason i have it.

New comment by johnxianren in "Show HN: The load-bearing vocabulary of Claude"

johnxianren — Fri, 28 Aug 2026 07:28:08 +0000

It's not picking the word, the harness literally told it to plant it.

New comment by johnxianren in "My Business Is Dying"

johnxianren — Fri, 28 Aug 2026 07:14:17 +0000

The hell part is the delay. Says the PR is done, but you only see the cliff once you read it.

New comment by johnxianren in "Show HN: Automatically detect and patch walking-dead states in Sierra games"

johnxianren — Wed, 19 Aug 2026 08:47:53 +0000

The delay is the real killer. You don't realize you're driving off a cliff until you’ve already hit the ground.

New comment by johnxianren in "AI usage patterns in software teams"

johnxianren — Wed, 19 Aug 2026 04:01:00 +0000

Felt that. I already gave DeepSeek the HTML and it still said the UI was good to go, then told me it has no vision.

New comment by johnxianren in "Agent skills should be compiled, not just read"

johnxianren — Mon, 17 Aug 2026 16:04:57 +0000

Does it though? From the writeup it just looks like it pulls out the steps and gates. You're still authoring one SKILL.md either way.

New comment by johnxianren in "Agent skills should be compiled, not just read"

johnxianren — Mon, 17 Aug 2026 15:46:55 +0000

Enforcing steps is fine, but mixing the orchestration layer into the methodology doc is just a mess. Skills should probably stay separate anyway, just keep instructions short

New comment by johnxianren in "AI companies are shredding rare books"

johnxianren — Mon, 27 Jul 2026 13:07:26 +0000

I have zero proof for this, but just a what if: what if Anthropic's strict anti-China stance actually means the Chinese training corpus is way more valuable than people realize?

New comment by johnxianren in "Why Software Factories Fail (or: harness engineering is not enough)"

johnxianren — Fri, 24 Jul 2026 09:57:59 +0000

So true.

So when you see that split coming, do you drop a quick warning, or just copy-paste and leave the mess for tomorrow?