Muratcan Koylan

It’s actually a good question; the difference is subtle but structural. I usually frame it like this: AGENTS[.]md acts as the declarative context. You write this for every repo (and nested directories) to define the project structure, persona, and coding rules. Skills are the functional protocols. They provide the agent with modular capabilities like advanced tool-use and multi-step chaining that are dynamically discovered only when needed. If AGENTS[.]md defines the identity and environment (the body), Skills provide the specialized toolset (the capabilities) used to execute tasks autonomously.

Is there a practical way to take a large open-weights LLM and make it more task-focused by removing irrelevant weights? The goal is a smaller, more efficient model that’s cheaper and faster to run on hosted inference. Pruning and knowledge distillation seem like the main approaches, but I’m not sure how much accuracy typically drops, especially with task-aware techniques that try to protect performance on the target task. I’m particularly interested in practical, low-effort solutions that I can actually test without spending too much time or money. Ideally something that delivers both: - Better task focus/specialization - Real cost savings on hosted inference (faster, cheaper to run) If you have a workflow, tool, suggestion, or technique that works well (especially ones that combine pruning with quantization, or task-specific pruning methods), please drop the details below.

He takes scamming to a whole new level

The prompt engineering here is super impressive! Such a great example of agent prompting:

We desperately need a smart model router. 1. We’re seeing a model explosion: GPT-5.6, Grok 4.5, Muse Spark 1.1, GLM-5.2, and Fable 5 all launched within the past month. 2. Even for a single model family like GPT-5.6, there're 3 (Sol, Terra, Luna) and 5 reasoning-effort levels. That is far too many decisions for users to make manually. The best model should be selected automatically based on the task, latency, quality, and cost.

Toronto’s no go zone — unless you want lungs full of fentanyl, crack or crystal meth

we're seeing supersized results with muse spark 1, comparable to fable when generating static sites at a much lower cost

The prompt engineering here is super impressive! Such a great example of agent prompting:

GPT-5.6 Sol Ultra produced a proof of a 50 year old math conjecture. Unlike the Erdős Unit Distance Problem, this was done with a model publicly available *today*. I look forward to seeing what scientists and researchers are able to do with this model!

While this is great prompting for the neutral observer, recognize that this comes from the expert mathematician who understands their domain and problem solving approaches and pitfalls. AI is an amplifier - feed signal, reduce noise.

New ChatGPT voice mode is incredibly impressive. I continue to believe that natural voice interface is underweighted in LLM assessments. For a decent chunk of people, it’s just how they interact with AI.