Does this mean the CLI generates a reusable script or abstraction for a specific site so it doesn't have to re-evaluate the entire page structure on every run?
[dead]
Show HN: Freu CLI – Cut web agent token usage by 90% via compiled browser skills
11–14 of 14 posts
Re: Show HN: Freu CLI – Cut web agent token usage by 90% via compiled browser skills
#12Does this mean the CLI generates a reusable script or abstraction for a specific site so it doesn't have to re-evaluate the entire page structure on every run?
Yes, and I tried the same labelling work with Kimi with the CLI, which costs me about $0.05, almost as good as GPT5.5, which costs $0.14.
Re: Show HN: Freu CLI – Cut web agent token usage by 90% via compiled browser skills
#13I haven’t tried it yet cuz setting up the Chrome extension feels kinda annoying ngl, but the idea is actually pretty exciting. I feel like this could solve 3 big problems: anyone can use it to make skills that are actually personalized to their own use case; token cost gets way lower for repeat stuff, which is basically what this repo is saying; a lot of businesses are scared to use agents for core workflows cuz they…
Yes, that is one way to eliminate hallucinations.
As for that specific solution, we might remove the Chrome extension later because we plan to use a local vision model for driving it.
Re: Show HN: Freu CLI – Cut web agent token usage by 90% via compiled browser skills
#14Does this mean the CLI generates a reusable script or abstraction for a specific site so it doesn't have to re-evaluate the entire page structure on every run?
Yes, and I tried the same labelling work with Kimi with the CLI, which costs me about $0.05, almost as good as GPT5.5, which costs $0.14.
Yes, there is a difference in the cost of early labeling work, and the execution costs later on will show an even greater gap. Our execution costs are near zero.