Often limited by "consecutive tool usage"

[maxekman]

The agentic workflow is great, but I'm often feeling limited by the tool usage (I'm a Pro subscriber using the provided Claude 4). It would be great to be able to allow it to continue with one more prompt and get additional tool usage (often without any additional input) instead of enabling Max/Burn mode, which will consume A LOT more prompts.

What do you all feel about this balance? My feeling is that the context window size limit is often ok, but the tool usage is limiting the ability, even for small/medium tasks across multiple files.

[hardbyte]

Agree, also a pro subscriber but in a monorepo I switch to the copilot versions of the same models because the zen ones barely gather the request context for any decent task before hiring tool limit.

[balboah]

yup, that was the first thing that I experienced when trying Zed for the first time as well. Did the newbie mistake of solving it by enabling burn mode, wiping out my 50 trial requests pretty quickly on the first task :)

[petri]

What is the rationale / purpose of consecutive tool use limit to begin with? Just a way to capture revenue for Zed?

If that's the case, whoever is the product manager or has the corresponding role should re-evaluate their revenue model. Making things harder via extra warnings & UI actions and hard to understand payment consequences do not make for a product customers love.

[maxekman]

Agreed! There must be some useful middle way between “a bit more tool use” and Burn mode. That would be lovely.