Issue Summary

When using Rovo Agent to access files (PDF, csv) attached to Confluence page, results are inconsistent. Sometimes, it can successfully extract the information from the file and analyse as per the agent's instructions, but sometimes, Rovo would respond that it doesn't have access to the file.

Steps to Reproduce

1. Attach a PDF file on a Confluence page
2. Specify the PDF file name to be used in the instructions of the Rovo Agent, along with the instruction to extract information from this specific file. 
3. Prompt the agent to tell about the contents of the file. 

Expected Results

Rovo Agent should be able to extract the information successfully, and return the information to the end user. 

Actual Results

The results are inconsistent. Sometimes, Rovo Agent returns the needed information correctly. At times, it says it doesn't have access to the specific PDF file and could not review its contents. 

Workaround

Provide the PDF file directly to Rovo Agent via the 'Upload File'