Tech News
Claude Code reads AGENTS.md only when telemetry is on [fixed]
A bug in Anthropic's Claude Code CLI caused it to read the AGENTS.md configuration file only when telemetry was enabled.
From portal-hopping to instant answers: HEMA’s journey with MCP and Amazon Bedrock
HEMA, a 100-year-old Dutch retailer, turned developer portal-hopping into instant answers by building HAL, an internal AI assistant on Amazon Bedrock AgentCore.
Ringg’s AI agents resolve up to 65% of customer calls with OpenAI
Using GPT-5.6, Ringg powers multilingual agents across voice, chat, WhatsApp, and web for 90% less cost.
How SWE-Serve Exposes the Gap Between Local Tests and Live Serving
An AI coding agent’s patch can pass tests yet fail when the server loads a real model and handles requests. Evaluating changes to inference-serving software.
Claude Opus 5.5, GPT-6 Sol, GPT-6 Luna, and a new price war
Yesterday was Grok 4.7 ( pelicans ) and MiMo v2.6 Flash/Pro ( more pelicans ). Today Anthropic released Claude Opus 5.5, and around an hour later OpenAI released GPT-6 Sol and GPT-6 Luna.
GitHub Repos
feder-cr/invisible_playwright_mcp
Your AI agent browses the web without captchas: self-hosted Python MCP server on undetected anti-detect stealth Firefox, for browser automation, scraping, computer use.
# Python# agentic ai# ai agentOpenHands/OpenHands
🙌 OpenHands: AI-Driven Development
# TypeScript# agent# artificial intelligenceEKKOLearnAI/ekko-studio
Ekko Studio is a local-first AI workspace for multi-agent chat, coding, and visual workflows, available on desktop and the web.
# TypeScript# agent# ai agentResearch Papers
Mizar: A 159M-Parameter Audio-Language Model for Audio Understanding
Mizar is a small 159M-parameter model that connects a compact audio encoder to a tiny language model so it can answer questions about sounds. It is trained in three stages and runs on a single CPU, beating other sub-200M audio-language models on three benchmarks.
Towards Efficient Reasoning: Learning Causal Shortcuts for Diffusion Language Models
This paper proposes a training method for diffusion language models that finds important token chains ('causal shortcuts') and masks them preferentially so the model learns to reach correct answers faster. It reports small but consistent accuracy gains over standard...
Reference-Based Analysis of Coherence and Diversity in Open-Ended Text Generation
A new framework evaluates open-ended text generation by comparing coherence and diversity to human-written continuations, using alignment, summaries, and likelihood under a human reference distribution.