Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
ai-tamer
searching Neon…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
8 ms
·
1.
▲
Coding agent is under-specified
(hsaghir.github.io)
2 points
by
ai-tamer
4mo ago
|
0 comments
2.
▲
by
ai-tamer
5mo ago
Same here. Spent 5 minutes blaming my VPN before HN saved me.
3.
▲
by
ai-tamer
5mo ago
We're rich ;-) More seriously, in a multi-agent setup the per-token cost matters less: a bit of Claude, a bit of Codex, a bit of Gemini-CLI, ... No single model carries the full bill, and having three different training sets catches m
4.
▲
by
ai-tamer
5mo ago
Good, but "Vibe" is a surprising name. A year ago Mistral Code was sold on engineering rigor and CISO comfort. "Vibe" evokes exactly the opposite, and to me the term has carried a dismissive ring in dev circles since Kar
5.
▲
by
ai-tamer
5mo ago
Same. The numbers match your feel. Going from 4.6 to 4.7: +14.6 on MCP-Atlas, +10.9 on SWE-bench Pro, tool errors cut by two-thirds. But BrowseComp dropped 4.7 points. Anthropic's own announcement says 4.7 "takes the instructions
6.
▲
by
ai-tamer
5mo ago
Zhao et al. 2023 showed any imperceptible watermark is provably removable by generative regeneration: pass the image through an img2img or VAE, the model reconstructs it visually identical but starts from a different latent. Watermark gone.
7.
▲
by
ai-tamer
5mo ago
(Disclosure: I work on tamer, an OSS supervisor for coding agents — biased.) Add one more to the count. The OAuth-across-harnesses idea would help, but it doesn't fix the shape of the problem. "Harness" has always felt off to
8.
▲
by
ai-tamer
5mo ago
Genuine question: how many agent-hours to rebuild Jira from scratch and migrate 100% of the content out? Split the work, pool our agents, ship by August 17. ;-)
9.
▲
by
ai-tamer
5mo ago
https://arxiv.org/abs/2604.04721 I like it when researchers confirm my intuition
10.
▲
by
ai-tamer
5mo ago
Note to self: don't go away for the weekend without HN ;-) Personally I run several agents. At minimum Codex and Claude, so they cross-check each other. Exactly to avoid what you describe. Duplicate functions or "tests all pass&qu
11.
▲
by
ai-tamer
5mo ago
Fully agree. Non-dev solutions are multiplying, but devs also need to get much more productive. I recently asked myself "how many prompts to rebuild Doom on Electron?" Working result on the third one. But, still buggy though. The
12.
▲
by
ai-tamer
5mo ago
Do you ask it for a design first? Depending on complexity I ask for a short design doc or a function signature + approach before any code, and only greenlight once it looks sane.
13.
▲
by
ai-tamer
5mo ago
It's annoying when a project you depend on gets taken in a direction you didn't sign up for. But the silver lining here is that the cost of maintaining a fork has dropped dramatically. A few years ago, forking something like 11ty