yruzin

5 ms·

1.
Show HN: Aplexica – resume a Claude Code conversation in Codex, and back(github.com)
1 points by yruzin 2h ago|0 comments
2.
End AI agent lock-in. Start conversation with one agent – continue with another(github.com)
1 points by yruzin 14d ago|0 comments
3.
Aplexica – keep memory and conversations when you leave Cursor for Claude(github.com)
1 points by yruzin 15d ago|0 comments
4.
What 50 open source projects taught us about security in the AI era(github.blog)
4 points by yruzin 1mo ago|1 comments
5.
Self-Evolving Coding Agents(arxiv.org)
2 points by yruzin 1mo ago|0 comments
6.
Tailscale Mitigates the Lethal Trifecta(tailscale.com)
3 points by yruzin 1mo ago|0 comments
7.
A Hermes Agent Skill Looping Between Codex and Claude Code(codenote.net)
1 points by yruzin 2mo ago|0 comments
8.
IssueTrojanBench: Benchmarking AI Coding Agents Against Malicious Issue Requests(arxiv.org)
2 points by yruzin 2mo ago|0 comments
9.
The Kotlin Benchmark for AI Coding Agents(blog.jetbrains.com)
2 points by yruzin 2mo ago|0 comments
10.
by yruzin 2mo ago
I think this is where harness makes a lot of sense. Use LLM to produce all possible attack angles/phrases and just stupidly filter them out on input.
11.
Orca-Bench: How Ready Are Language Model Agents for Oncall?(arxiv.org)
30 points by yruzin 2mo ago|11 comments
12.
The GitHub Platform Era Is Over(julien.danjou.info)
3 points by yruzin 2mo ago|0 comments
13.
The harness is all you need (mostly)(github.blog)
2 points by yruzin 2mo ago|0 comments
14.
No Snake Oil: Verifying Python Package Builds(arxiv.org)
3 points by yruzin 2mo ago|0 comments
15.
No-Look Coding and the Five Stages of Grief(blog.kindel.com)
4 points by yruzin 2mo ago|6 comments
16.
Just Say the Word(newsroom.spotify.com)
3 points by yruzin 2mo ago|0 comments
17.
Denying the Worm(crowdstrike.com)
2 points by yruzin 2mo ago|0 comments
18.
Malicious Agent Skills in the Wild(arxiv.org)
2 points by yruzin 2mo ago|0 comments
19.
The /Goal Function Made Loop Engineering Click for Me(bhart.org)
3 points by yruzin 2mo ago|0 comments
20.
A Scorecard for the AI Age(openai.com)
3 points by yruzin 2mo ago|0 comments
21.
Prompts and Skills need a system of record(mistral.ai)
2 points by yruzin 2mo ago|0 comments
22.
The Server Is a Sync Relay Now: Architecting Around Client-Owned State(tiarebalbi.com)
3 points by yruzin 2mo ago|0 comments
23.
Building scalable AI agents with modular prompt transpilation(developers.googleblog.com)
8 points by yruzin 2mo ago|2 comments
24.
by yruzin 2mo ago
Claude is Extremely slow, especially fast few days. Codex is so much faster in in my opinion with comparable quality. Anthropic is going to fix their problems, but for me as a user that depends for the service to work, it's not accepta
25.
by yruzin 2mo ago
I pretty much giving up on Claude. It's Extremely slow, especially fast few days. Computer use takes so long to do anything that sessions expire. I switched to Codex and it's a different story. I'm not saying that Anthropic
26.
by yruzin 2mo ago
How did you measure that the agent used >90% of time re-reading the context? The idea is that the agent has the KV cache, that helps with inference, but I'm not sure if what you are saying is related to re-doing the inference or the
27.
by yruzin 2mo ago
Sorry, my bad :)
28.
by yruzin 2mo ago
Spoiler for the quiz below. The defer delays the function call, but it does not delay evaluation of the call’s arguments. This means that both timestamps are therefore captured near the beginning of the request, even though the second line
29.
by yruzin 2mo ago
Durability is not the problem, but because the state lives somewhere you structurally can't relocate. That constraint never shows up in a benchmark, it shows up in an enterprise procurement call.