yzhong94

If you run it with a long running sandbox then yes

Not just system prompts. Everything on Broccoli goes through multiple rounds of design critiques, and then multiple rounds of code review

We have not! How much control do you have over Cursor Cloud Agent?

I'd say it's one-shot for 60% of the PRs we ship today. As I mentioned in the post, the rest still go through a lot of back and forth especially at the design stage.

I think it totally depends on the team on what the correct workflow is. The agentic workflow should accelerate what you have instead of asking your team to adopt something new. This is why we think it helps to own the orchestration yourself so you have that level of control!

Thank you! We found this approach for both plan critique and code review to be extremely effective

Haha we are definitely like-minded because our internal Broccoli is actually on Firestore. That being said, Firestore is an acquired taste so we rewrote the OSS backend to Postgres so that everything can be deployed in one go with the infra that people are most familiar with.

Re: spot VMs. Great idea! There are two features we have not finished porting to OSS. Internally, we can specify the instance type and timeout, and we also send about 50% of jobs to Blaxel; we find it has a much better cold start compared to Cloud Run. We probably will port the multi-vendor support logic over to OSS soon but wanted to keep v1 simple (and a one-provider magic experience!).

Scheduler is a wish item for us. Curious how you implemented it? Currently, we just have a scheduled Cloud Function during the night to automatically address open PR comments (via the Broccoli GitHub feedback automation) so that the engineer wakes up to a mostly clean PR without needing to do anything. We haven't ported this to the OSS yet because 1) Firebase Cloud Functions, 2) not sure what would be the best ergonomics. Any suggestions here?

We use firebase which supports the preview environments. It's mainly for front end changes though. Are you looking for a solution for backend changes as well?

100% on using linear as the source of truth! We went a bit further and just use Linear as the prompt. So there isn't a human written one adding context that lives elsewhere

We also have a multi-repo setup, to trigger it you can just tag two repos in the Linear label!

We don’t believe PM or eng can write the best prompt or spec, so we don’t ask them to.

One real Linear ticket from a few months back that we assigned to broccoli:

Store post-processing run outcomes in a versioned, append-only audit trail so re-running the same processor on the same audio file produces a complete history (who/when/what changed), while keeping an easy “latest result” view. Add an admin-only UI.

That’s it. As a part of the sketch step, broccoli does its own repo discovery and online research before planning the execution.

We are not trying to compete with OpenAI and Anthropic! We open source it because there's interest from other startups.

Teams would use Anthropic and OpenAI, but they shouldn't just use Anthropic or OpenAI. We see much better results from calling the models independently and do adversarial review and response.

This doesn't replace your need for the models, but you certainly don't need to rely on any of the cloud agent solutions out there that call these models underneath the hood.

The main difference is that you have full control over this!

let us know if you have any feedback!

Oh got it! In this case, the main difference is that we go through a flow from design to implement using our own prompts, and uses both Codex and Claude Code so they can improve off of each other.