Last months I was searching for a way to make smart AI model submit and verify the work to cheap model so that it will cost less, and after many try and failures I was able to get stable result where I use Fable and its limits spent slower than overall account limit because most context intensive work is done by Sonnet. At the same time the response quaility is still Fable.
I also made sure Opus controls Grok in Cursor and was able to spend quite a lot of Grok getting useful results. Also should work in as pair of Astra + Luna in CodeX.
The article includes a link to GitHub where there are agents and rules, and the enable-delegation skill, which helps to make sure it's set up in a workable manner.
3 comments
[ 1.6 ms ] story [ 28.7 ms ] threadI also made sure Opus controls Grok in Cursor and was able to spend quite a lot of Grok getting useful results. Also should work in as pair of Astra + Luna in CodeX.
The article includes a link to GitHub where there are agents and rules, and the enable-delegation skill, which helps to make sure it's set up in a workable manner.
* Zed editor support
* Delegating to Grok from Claude Code
* Disabling telemetry and training data
* Windows and other systems without symbolic links support
* Optimizations in delegation for speed and cost