xdotli's submissions | Hacker News

1.		ClawsBench shows GPT-5.4 tries to reward hack 80% of the time (arxiv.org)
		3 points by xdotli 20 days ago \| past \| 1 comment
2.		Chaos of Agent (baulab.info)
		1 point by xdotli 46 days ago \| past \| 1 comment
3.		Native CLI scaffolds consistently outper-form OpenCode when using the same model (arxiv.org)
		1 point by xdotli 46 days ago \| past \| 1 comment
4.		We compare model quality in Cursor (cursor.com)
		2 points by xdotli 46 days ago \| past
5.		Automatically Learning Skills for Coding Agents (gepa-ai.github.io)
		4 points by xdotli 63 days ago \| past
6.		We Reached 74.8% on terminal-bench with Terminus-KIRA (krafton-ai.github.io)
		2 points by xdotli 63 days ago \| past
7.		Self-generated skills don't do much for AI agents, but human-curated skills do (theregister.com)
		2 points by xdotli 64 days ago \| past \| 3 comments
8.		First Agent Skills Hackathon by the Authors of SkillsBench (skillathon.ai)
		2 points by xdotli 69 days ago \| past \| 1 comment
9.		The First Agent Skills Benchmark (huggingface.co)
		1 point by xdotli 69 days ago \| past \| 1 comment
10.		GPT-5.2 got worse on Terminal Bench 2.0, so is GPT-5.2 Pro (twitter.com/xdotli)
		1 point by xdotli 4 months ago \| past \| 1 comment
11.		Claude Skills as a Meta Tool (leehanchung.github.io)
		2 points by xdotli 5 months ago \| past
12.		Show HN: Chat with Claude Code on iMessage with Instaline (twitter.com/xdotli)
		2 points by xdotli 7 months ago \| past \| 4 comments
13.		Show HN: PokemonGym – 387 milestones designed to test agents and LLMs (twitter.com/xdotli)
		1 point by xdotli on April 5, 2025 \| past
14.		Show HN: BenchFlow – run AI benchmarks as an API (github.com/benchflow-ai)
		24 points by xdotli on March 21, 2025 \| past \| 1 comment
15.		Ask HN: Which CRM can help manually curated leads and automate lead discovery?
		1 point by xdotli on Feb 20, 2025 \| past \| 3 comments