7 comments

[ 5.8 ms ] story [ 35.7 ms ] thread
[flagged]
Seems interesting, but buried under a mountain of written slop.
It is marketing for some SaaS thing. However, I think something like this, "verification loops with some sort of tricks", is what Anthropic and "OpenAI" do behind the scene with their interfaces.
I told DeepSeek to read the article and investigate links and terms in the article.

The gist of it is that uses a bunch of skills from a solution called ironbee in a loop iterate with LLMs to help them solve the tasks better by improving the feedback loop via browser-testing, debugging, visual-testing, etc..

The skills can be found here:

https://github.com/ironbee-ai/ironbee-devtools-skills/tree/m...

Question that was left unanswered:

How much longer does a 4x DeepSkeep loop took compared to Opus?

From what I understood, the gist of it is that uses a bunch of skills in a loop to help LLMs solve the tasks better via browser-testing, debugging, visual-testing, etc..

The skills can be found here:

https://github.com/ironbee-ai/ironbee-devtools-skills/tree/m...

Question that was left unanswered:

How much longer does a 4x DeepSkeep loop took compared to Opus?

If the agent never sees the hidden tests, what is the verifier actually checking against? Trying to understand what "verification" means here concretely.