Very hard to find AI benchmarks that don't look like this (and yes, that includes obscure benchmarks that nobody would train on and benchmarks with holdout datasets)

openai.com/index/harnes... i have been mildly obsessed with the ideas in here for the past month or so. incredibly validating, total victory: my suspicions are correct, actually. "This functions like garbage collection" is a sentence I have said out loud. craziness

resources.anthropic.com/hubfs/The-Co... "What you'll get out of this guide: By the end, you'll be able to build a functional skill in a single sitting. Expect about 15-30 minutes to build and test your first working skill using the skill-creator"

It prevents wasted effort, keeps me in control of architecture decisions, and produces significantly better results with minimal token usage than jumping straight to code." boristane.com/blog/how-i-u...

Time to consider not just human visitors, but to treat agents as first-class citizens. Cloudflare’s network now supports real-time content conversion to Markdown at the source using content negotiation headers. https://cfl.re/4ksZQ1S

Anything you can do in Obsidian you can do from the command line. Obsidian CLI is now available in 1.12 (early access).

I wrote about the most ambitious form of AI-assisted software development I've seen yet - Strong DM's "Software Factory" approach, where two of the guiding principles are "Code must not be written by humans" and "Code must not be reviewed by humans" simonwillison.net/2026/Feb/7/s...

what you want is an MCP. oh wait have you tried installing a skill? no you don't have an agents,md file. actually what you want is a plugin. have you tried agent orchestration? you have gotta try hooks.

”limited-time $50 (USD, or local currency equivalent) in extra usage to Pro and Max users to coincide with the launch of Claude Opus 4.6.”

🚀 VS Code is becoming your home for multi-agent development. A single place to view and manage all your agents - local, background and cloud, and even Claude or Codex, all under your Copilot subscription. Learn more ➡️ aka.ms/VSCode/109blog

is it just me or is it getting hot in here?

Lot of good takes on this video: what has changed since December starting with big frontier model releases, Claude Code's Task System & parallel agents, the Capability Overhang, where bottlenecks are moving..

You told us you’re running multiple AI agents and wanted a better UX. We listened and shipped it! Here’s what’s new: 🗂️ Unified agent sessions workspace for local, background, and cloud agents 🔀 Parallel subagents 💻 Claude and Codex support for local and cloud agents 🌐 Integrated browser & more...

OpenBMB's MiniCPM-o 4.5 (open-weight) - Full-duplex Omni-modal Live Streaming: The model can see, listen, and speak simultaneously in a real-time conversation without mutual blocking - Proactive Interaction: Moving beyond reactive QA to performing proactive interaction, such as initiating reminders

Google presents PaperBanana Automating Academic Illustration for AI Scientists huggingface.co/papers/2601....

indydevdan video on Claude Code new tsk mgmt sys: www.youtube.com/watch?v=4_2j... neat: - coordinator agent uses templates to create custom builder + validator agent pairs for each task. - hook that runs a .py file that verifies the created plans always contains the parts defined in the template.

Claude Sonnect 5 "Fennec" speculation/rumors on reddit/x: - possibly February 3, 2026 release window. - Aggressive Pricing: Rumored to be 50% cheaper than Claude Opus 4.5 while outperforming it across metrics. - 1M token context window - 80.9% on SWE-Bench

hypercontext.sh : Hypercontext is a knowledge skill for Claude Code. Invoke /hypercontext and the agent introspects — threads, files, tools, remaining context — and renders a spatial ASCII map.

A2UI: Agent-to-User Interface

Introducing the Letta Code SDK: the open source alternative to the Claude Agent SDK - memory-first design - open source & model agnostic Combine with ChatGPT Codex, Zai, MiniMax coding plans. You can also use your Letta API credits for any other model.

Anthropic prepares new sketch tool, Cowork in projects and more

MCP Apps is getting support from many sides now Anthropic, OpenAI, vscode… interesting to see how gen-ui and things like AG-UI/CopilotKit bring to table

Today, the MCP community is announcing MCP Apps, the first official MCP extension. VS Code is the first major AI code editor with full MCP Apps support. With MCP Apps, tool calls can now return interactive UI components that render directly in the conversation. Learn more: aka.ms/MCPApps/Blog

Claude Code TeammateTool gist.github.com/kieranklaass...