Been facing a few issues, which I am not sure if anyone has found a solution to these yet.
I’ve been scanning around for an answer but so far, I have not been able to crack this. Please could you help?
1. Correct agents not being selected for the right job >> “build_from_library”
I’ve created my own list of agents with clear descriptions. Descriptions are one-liners that do not overlap
e.g. You are a Data Analyst, a highly specialised in analysing all types of data, numerical, text-based, natural language
2. Agents not following their own system_messages
They keep doing things that is not in their remit. e.g. a coder will try to analyse data despite being explicitly not to in their system_message and despite not having the required tools AND a data_analyst will try code
3. Agents not fully completing the required task
e…g message = Each agent should confirm their understanding of these instructions and their respective roles in the context of the end solution
4. Create workflows, where I’ve pre-committed agents for specific use cases
I’d provide the instructions/rules