No matter what I do, it will generate something out of context, its loading my documents, reading them but not using them. So annoying.
Matching tool name. Make sure the tool name is exactly prompteer_knowledge_search, otherwise the agent won’t recognize the tool.
In the prompt, you should be clearer and avoid using “.” as the initial input; give the agent more specific instructions.
The agent should automatically call the vector store to retrieve context from the vectorized documents. The model’s response should incorporate real fragments of the content, maintaining consistency with the instructed tone and style.
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.