Claude Mythos is suspected to be a Recurrent-Depth Transformer (RDT) — also called a Looped Transformer (LT). Rather than stacking hundreds of unique layers, a subset of layers is recycled and run through multiple times per forward pass. Same weights. More loops. Deeper thinking.
这一观点挑战了传统大模型架构的常识，认为Claude Mythos的核心创新不在于增加参数量，而在于通过循环使用相同权重来实现更深层次的推理。这种架构设计反直觉地表明，模型的'深度'可以通过循环迭代而非堆叠层来实现，这可能解释了Mythos在复杂推理任务上的优异表现。
The core philosophy of MCP is simple: it's an API abstraction. The LLM doesn't need to understand the _how_; it just needs to know the _what_.
这是一个深刻的架构洞见，揭示了MCP与Skills的根本区别。MCP通过抽象API实现了关注点分离，使LLM只需关注'做什么'而非'怎么做'，这种设计大大简化了AI与服务的交互复杂度，代表了更优雅的工程思维。
Invisible architectures, on the other hand, consist of ways that our worldis structured for us and by us, except that the structure itself is not imme-diately visible
for - meme - invisible architecture - insight - language - as unintentional invisible architecture