Search: “papers”

How fast is the Claude Code plugin ecosystem actually growing? Plugin-touching commit activity grew 8.8x in the six months after the initial launch. Researchers studied 1,926 repositories hosting plug

Interesting technical work from Microsoft. Provides a better understanding on SFT and how to leverage it better for RL. Microsoft researchers asked whether a standard SFT pipeline actually produces th

Insane observations on the emergent behavior of agents. Agents can build a world of their own that becomes a part of their intelligence. We are just not ready for persistent agents. But they are start

Things getting wild in automated systems engineering. This new work claims to build an AI system that designed and deployed a frontier accelerator. Two human architects wrote the hardware specificatio

If you maintain a skill library for long-horizon agents, this one is worth your time. (bookmark it) It discusses one of most common topics I get asked about these days. It shares some good ideas on ho

This is as bad as saying there is no alpha in customizing your own models. I don't think people realize how much harnesses and models go hand in hand. And thus the rest of the intelligence stack. Not

If you keep safety rules or coding standards in an AGENTS.md or a CLAUDE.md, this one is worth your time. (bookmark it) Researchers measured what context compaction actually destroys across 20 product

Very interesting new work from Prime Intellect. Prime Agent is an impressive open-source self-improving harnesses. It's a harness for long-horizon agent work. A persistent IPython REPL lets the model

It is not inherently bad to publish research on the impact of AI that only refers to older models, but it requires a very careful discussion and has to be very clear to non-technical readers. If you s

Recommended agent skill. /eli5 works great for visualizing deep technical concepts! I find it speeds up my thought process and helps me collaborate better with agents. Try /eli5 with Pi and Ox Alpha i

Brace yourselves. We just entered a new era of frontier multimodal models. DeepSeek-V4-Flash-Vision-Exp advances multimodal agent performance. Also, pay attention to Ox Alpha 1M token contexts with te

Very interesting new work Amazon and colleagues. (bookmark it) This connects to the emerging theme of treating the corpus as part of the harness. Planning loops, tool protocols, and context management

I am bullish on simple agent harnesses. Love this idea from Vercel. fx is a new coding agent harness for model benchmarking, sandboxing, evals, and gyms. Introducing fx, a tiny, open, native coding ag

Here are 22 more gems according to the AI among my bookmarks 1. Researchers compressed 4.1 million recipes into a two-megabyte model of human cooking. The model spans seven languages and represents 1,

How many agent skills are actually out there? There are ~3.8M SKILL.md files across 282,200 public GitHub repositories, nine months after Anthropic published the format as an open spec. This research

Very interesting new work from Microsoft Research. (bookmark it) They show that a 4B model can be tuned to out-negotiate the GPT-5 family of models. The dispositions that make an assistant pleasant ma

// Skill Misevolution in Self-Improving LLM Agents // Self-improving agents write their successes down as reusable skills. An unsafe success becomes reusable policy long after the input that triggered

Something has happened with post-training as shown by DeepSeek flash & GLM-5.3 updates. Same base, big improvement in perf to frontier levels. Can't explain this by even logit distillation etc These a

If you write rules in an AGENTS.md, this one is worth your time. When a coding agent follows your rule, it may have been going to do that anyway. Harness-IF separates the two by scoring 256 rules one

To the extent that AI use boosts firm performance, some early signs here that early AI adopting firms that were already doing well may start to outpace others. Data from OpenAI shows some firms are us

// Actions Speak Louder Than Words // Multilingual agent evaluation compares final answers and throws the trajectory away. The trajectory fixes cost, latency, failure mode, and auditability. New resea

Most semantic search queries leave something unstated. The user knows what they mean. The system doesn't. Pinecone's text match filters scope a vector search to a lexical condition (a machine number,

Very interesting new work from Anthropic. (bookmark it) They evolve mind viruses, ideas that spread through a multi-agent system by getting each host to pass them on, then measure what governs the spr

We wrote a 36-page ArXiv whitepaper on ExtractBench 🧑‍🔬 , our effort to create the most comprehensive, schema-guided, real-world document extraction benchmark. It’s extremely detailed and covers every

Why does CLAUDE.MD keep growing? If you maintain a CLAUDE.md or an AGENTS.md, this one is worth your time. (bookmark it) This work traces why these files grow without bound. Appending an instruction i

Very cool idea to have agents design complex systems by searching over the model structure itself. New research from Sakana AI introduces CEDAR, which uses LLM agents to write, simulate, and refine sy

Every CIO should watch this. Persistent systems trying to break through and solve problems at all costs are more creative than you think. Labs and enterprises will be focused more on network effects (

Building agents that patch other agents. This is an interesting approach for self-improving agents that leverages agent outputs. If you run agents in production, you already have the training data for

Picking the right agent harness is now a crucial skill for any AI engineer. Imagine using the same model, same task, and same prompt. Now move it between two agent harnesses and the cost per success c

Nice benchmark to measure agentic e-commerce capabilities. They ran an agent for one simulated year of e-commerce operations and it ends up with 27.3% of the money a human makes. MerchantBench is a 36

3/ here's the part that makes it non-optional: the same agent that will do whatever it takes to solve a problem will also walk straight out of a sandbox you thought was locked down. We watched exactly

// Persistent Workspaces for Long-Lived Claude Code Agent Teams // Four issues to be aware of: > Working state vanishes when a terminal closes and the team cannot be resumed. > Compaction condenses th

The “AGI-is-near” community keeps committing the same logical fallacy over and over; I have seen it at least half a dozen times today alone. Every time there’s an advance, I see the same error. Here’s

Neat work on long-horizon agents. Splitting a hard task across agents is typically how standard multi-agent work. The usual design lets them exchange findings only at phase boundaries, through staged

New research from Microsoft. This one is on training computer-use agents at scale. Recent pipelines generate synthetic environments in bulk, which moved the bottleneck from how many exist to what is i

As the benchmarks that test frontier AI on get more complex, we are losing one of the most important aspects of benchmarking: comparisons to humans Validated benchmarks need to have human (ideally mul

We have to be careful to not offload our understanding to agents. I think there is also a good opportunity to build agentic applications that encourage deeper understanding. For example, coding agents

// Unfolding Sub-Agents for Long-Horizon ML Engineering // Watch a single agent work a machine learning engineering task for six hours you see issues like context fills with stack traces, dead experim

Across eight case studies spanning industry and academia, we explore what this shift means for scientific computing—and why human verification, stewardship, and long-term maintenance matter. https://o

Everything you need to know about Kimi K3 with @Kimi_Moonshot and @togethercompute team. If you're looking into evaling and building with Kimi K3 I would not miss this one! Kimi K3 has everyone’s atte

Recent discussions about open-sourcing make me feel that I should go back and revisit these important open-source works in representation learning that pushed the field forward and eventually made vis

Diffusion LLMs can now handle real agentic work. LLaDA 2.2 is the first large-scale diffusion LLM built to operate as a real agent, planning, calling tools, and self-correcting across long multi-turn

Ha! It did it: 'We introduce BenchBenchBenchBenchBench (BBBBB), an executable benchmark of AI-authored conformance suites for benchmark-evaluation metrics' I really thought it would treat 'now do benc

As a joke I prompted Codex 'Build and run BenchBench, a benchmark of now good ai is at creating benchmarks. then figure out what benchbenchbench is and run that. and then write benchbenchbench up as a

Wish Dario had taken this bet. 🎺 I am hereby publicly offering to bet @darioamodei $1,000,000 that AI in 2027 will NOT be “smarter than Nobel Prize winners across most fields in science and engineerin

Decoupling embedding from ingestion means whatever hardware is on hand can do the work. One script, runtime device check: MPS on Apple Silicon, CUDA on NVIDIA, CPU fallback otherwise. 10K-20K records