Rovo "Input Too Large"

[Chris Green]

Am I trying to overwork Rovo?!? :) 

I have an agent I created to help with root cause analysis over a time period (i.e. prior month or quarter). I'm still in the process of testing and refining my prompt and output, and I'm more often than not getting "input too large". Weird thing is, I can set it aside and come back in a few hours and then it works just fine. Any thoughts here? Am I correct in thinking that my agent instructions might just be too large and Rovo is timing out? 

[Arkadiusz Wroblewski · Accepted solution]

Hello @Chris Green

Well, hell yes. Definitely. When you look into the community forum, you will find a lot of information about too long requests causing problems.

Prompts should be Paginated where you can, you know Rovo is still developing. We need quite some time till we will have predictable behavior.

Best,

Arek 🤠 

[Chris Green]

"Well, hell yes." - best response ever! :) 

Happy Friday! 

[Arkadiusz Wroblewski]

I mean. Answer should be true and authentic.

Rovo or chat GPT will answer you: You must be full aware that your prompt length have influence on Successful action etc etc 🤣 Still right. But a little soulless 😂😅

By the way, people are complaining about Rovo using data, but Chat GPT is on one side. Rovo has access to a huge library of a very specific market segment, and I'm curious about future development.

Happy Friday 🤠☺️

[Dr Valeri Colon _Connect Centric_]

@Chris Green you may be hitting a context limit, but the agent instructions alone may not be the cause. A quarterly RCA can accumulate Jira results, comments, history, and tool output. Try smaller time periods in fresh chats and compare results. For larger analyses, split the work by period/category or use subagents, then synthesize the results. Since it sometimes works later unchanged, capture failed request IDs for Support too.