Show HN: We analyzed 1,573 Claude Code sessions to see how AI agents work
Also: I open-sourced the behavioral ruleset and toolkit I built after 3,667 commits with Claude Code -- 56 slash commands, 318 skills, 23 agents, and 9 rules that actually change how the agent behaves • I got tired of managing Claude Code across multiple repos, so I built an open-source command center for it — with an orchestrator agent that controls them all
Show HN: OneCLI – Vault for AI Agents in Rust
Document poisoning in RAG systems: How attackers corrupt AI's sources
Meta Delays Rollout ofNew A.I. ModelAfter Performance Concerns - TheNewYork Times
Also: Meta Delays Rollout ofNew A.I. ModelAfter Performance Concerns - TheNewYork Times • Meta delays rollout ofnew AI model'Avocado' amid performance concerns, NYT reports
Claude now creates interactive charts, diagrams and visualizations
Also: Claude now creates interactive charts, diagrams and visualizations
Innocent woman jailed after being misidentified using AI facial recognition
Also: Innocent woman jailed after being misidentified using AI facial recognition
Executing programs inside transformers with exponentially faster inference
I spent 8+ hours benchmarking every MoE backend for Qwen3.5-397B NVFP4 on 4x RTX PRO 6000 (SM120). Here's what I found.
Also: I spent 8+ hours benchmarking every MoE backend for Qwen3.5-397B NVFP4 on 4x RTX PRO 6000 (SM120). Here's what I found.