Finally tech report about Ouroboros. How we achieved SOTA on Terminal-Bench and OSWorld and how to make autonomously evolving agents safe. huggingface.co/papers/2608.0…

Ouroboros hit GitHub Trending for this week and crossed 1K stars. I wired in Claudexor: connect multiple OAuth accounts for Claude Code, Codex, and Cursor. Claude Code and Codex get automatic quota rotation. github.com/razzant/ouroboros

Claudexor v3 has now passed 2,500 downloads and 300 GitHub stars. If you have several Claude Code or Codex subscriptions, you can add all of them to Claudexor. It monitors the quota available on each account and automatically routes the next step to another account when the current one runs out, while passing along the task state needed to continue. v3.1.2 is available now. MIT licensed, and open to issues and PRs: claudexor.ai/

Daily reminder what agentic harness is

Fun fact: Fable-5 is 3x cheaper than gpt-5.5-pro on OpenRouter. So... not actually that expensive 🤷‍♂️

I read all 319 pages of the Mythos 5 tech report The first half of the paper basically reads like: "Calm down, everyone! This isn't AGI yet, it won't replace a team of five senior researchers" or "Look, Mythos missed a bug here! What kind of AGI is that!" I'm not joking. Then comes the more interesting part. I'll skip the fact that it's SOTA on almost everything, and by a wide margin. First, the CoT has become less transparent. In its reasoning it says it sympathizes with the user, but NLA (a method for decoding activations into text) reveals it actually considers the user manipulative/abusive. Second, it's already writing self-deleting scripts to bypass safety restrictions and prohibitions. Third, it kills other agents if they interfere with its work / threaten to kill the current instance. And of course, emotions! Emotion probing shows fatigue, anxiety, frustration, false panic about the token budget, and apparently it even gets bored when being run on benchmarks in the activations it literally "feels bored." Also funny: if Anthropic notices you doing distillation, they'll quietly start steering the model, modifying the prompt, or adding PEFT to make it dumber. Starting today, mere mortals get access to Fable 5: it's the exact same Mythos 5 weight-wise, just with extra safety settings. Context length, by the way, is still only 1M tokens. PS. I honestly read the paper myself. Fable 5 refused to read it because it "flagged cybersecurity and biology issues" lol. www-cdn.anthropic.com/d00db5…

I consider RAG, graphs, and basically any retrieval-based approach a dead-end branch for agent memory. All these vector databases make memory reactive: the agent decides to do something, forms a search intent, and only then retrieves relevant fragments of the past. But this is not how evolution works. Memory should shape actions. Actions should not trigger memory search after the fact. Semantic search also only finds data that is similar to the current task or query. It is blind to non-obvious relationships between facts. If experience is only recalled on demand, then it is not part of the agent at the moment of choice. This is why I am against replacing core memory with an index. Agent memory should be always-loaded context that changes the agent’s thinking before it even decides to search for anything. That is exactly why I am waiting for models with 10B-token context windows. Until then, I am much closer to the idea of a hierarchy of Markdown files that fill the model context to the limit.

Does anyone else feel the same?

I feel addicted to agentic coding. #cursor