StErMi

#web3 dev + auditor | @SpearbitDAO LSR, @immunefi bug hunter, sage of AAVE codebase :D

[StErMi @StErMi]

During the next few days, I will share some of my private security research work that I have done in the last year.
​
All those projects are @aave related, and I feel very proud to have been chosen as one of the security partners to review them. I'm pretty sure that the dedication and knowledge that I have demonstrated on their codebase has been a pretty good investment 😁
​
Those projects were very challenging, but at the same time it was fun and a pleasure because you always feel motivated and engaged when the quality of documentation, specifications and code is high quality.
​
When @avara or @bgdlabs contacts you, you know that you are going to enjoy it. They are both a top-notch team to work with.
​
The first one that I want to share is a.DI (Aave Delivery Infrastructure)

It's a cross-chain communication abstraction layer for decentralised systems like the Aave DAO to communicate across networks, minimising the risk of underlying individual bridge provider failures, via consensus rules.

[StErMi @StErMi]

@every will "Thesis: 2027" be uploaded to YouTube so I can save my progress and have custom options like speed multiplier?

[StErMi @StErMi]

@nisedo They should enable it to individuals

[StErMi @StErMi]

@OpenAI is Daybreak Red restricted to companies or also indivisuals can apply to it?

[OpenAI @OpenAI]

Introducing GPT‑6 Cyber: a major advance in security research that pairs stronger capabilities with better model alignment.

The model powered the Linux kernel research we announced yesterday, and security teams at companies like Mozilla, Palo Alto Networks, Calif, and Cisco—alongside our own at OpenAI—are already putting it to work.

GPT-6 Cyber is available to select Daybreak Red customers and we are expanding access over time.

https://t.co/Nzab3Z1Td6

[StErMi @StErMi]

@clintgibler I assume that this is not based on how Codex Security (cloud/local) works right now right?

[StErMi @StErMi]

@OpenAI is you start a Codex Security Cloud Scan, by default it uses GPT-5.6 Sol with Daybreak Blue.

But as far as I get, GPT-6 Sol is also available for Daybreak Blue users (see https://t.co/1fQcNI3JL2)

Am I missing something? I would assume that GPT-6 Sol is more capable and cheaper (half the cost) compared to GPT-5.6 Sol

[StErMi @StErMi]

@0xProbabillity can I get the TestFlight invite?

[StErMi @StErMi]

@OpenAI please enable GPT-6 Astra to Daybreak Blue users. It's becoming impossible to do my work

[StErMi @StErMi]

@Dimillian is this what I was asking for, make side chats first-class citizens and "merge them" back into the main conversation? 👀

[StErMi @StErMi]

@finnvoorhees how can I access the TestFlight? 👀

[StErMi @StErMi]

@OpenAIDevs Is this not available for EU users?

[StErMi @StErMi]

@Dimillian It's stunning. Adding the glowing Apple logo is only by itself a reason to upgrade like when they finally added matte black as a color 😁

but I think that I'll wait for the M7 pro (next year?)

[StErMi @StErMi]

@AnaArsonist Thanks for the reminder! Started yesterday 😁

[StErMi @StErMi]

Am I wrong in thinking that we are at a point where we need to define a new standard around the "task/capability" concept? Both skills and subagents are "capabilities" that can be composed and enhanced (by providing access to other subagents, skills and tools) and the main difference is "how" they should be executed (fresh context? Inherit history? Being able to work in parallel?).

I feel that, at least as a solo developer, it's getting harder and harder to properly define the "task" (as written words) and then start to experiment how to "distribute" that task as a "sharable" thing. The same task can be executed as a skill, an agent or a subagent depending on what the specific part of the workflow need.

[StErMi @StErMi]

@m4rio_eth I think that those models are for companies that needs to do work that does not need the compute and reasoning need by the Astra level of models which are much more expensive.

[StErMi @StErMi]

@Zaddyzaddy @BugBunny_ai Does it support daybreak blue?

[StErMi @StErMi]

is there a reason why the default Codex behavior to "spawn" agents is with `fork_turns=all` that makes the subagents inherit the whole parent history inherited?

Isn't the whole point of subagents to start with a "fresh" context?

Am I missing something @OpenAI ?

[StErMi @StErMi]

@emilkowalski in the https://t.co/cpE24hUMB2 FAQ, you have the question "I’m a designer, not an engineer. Is this for me?".

You are missing the opposite one: "I’m an engineer, not a designer. Is this for me?"

[StErMi @StErMi]

I was thinking about the blog post "Building agents: When to fold subagents into your main agent loop" from @AnthropicAI this morning.

I think that we have settled on the best practice (does it really exist one in this field?) that you have to structure your "agent workflow" depending on the context and the needs and the eval result.

But being able to easily test it, evaluate it and maintain it has a cost and the same capability can't be easily embedded and provided as a tool, skill or subagent.

I feel like we are at a point that we need a new standard that enables us to define capabilities and then "distribute them" to the agents that can consume them as skills or subagents depending on the context/need we have defined for that agent.

https://t.co/miK46sEbC6

[StErMi @StErMi]

@cherry_mx_reds Is single-agent-loop an official @AnthropicAI skill? they have not published on https://t.co/JId2Ndw1lb as far as I see