Tool for batching tool calls?

[noctuid]

To upvote this issue, give it a thumbs up. See this list for the most upvoted issues.

Opencode has a batch tool that I've heard people mention they like for reducing requests. IIUC the supposed benefit is that it's easier to prompt models to use this batch tool then to prompt them to make parallel tool calls. I'm kind of skeptical of this approach but am not sure how to test parallel tool calling with eca. Is there any easy way currently to check whether tool calls are part of the same request?

[zikajk]

I liked the idea @noctuid , but then I ran into quite a few people complaining that it led to bad results. So, I don't know...

Maybe this bypasses the situation where the LLM avoids making parallel calls due to dependencies? 😅