#141791 - Better Error Handling

[AleksanderBak]

🔍 What's the problem?

Currently, the agent returns only the message "Internal server occurred" in case of an error. This is the only error type that makes it difficult to identify the root cause of the issue without analyzing the server logs. As a result, there is a need to expand the range of supported error types so that the user can determine whether the issue is temporary (e.g., Anthropic API overloaded - a common error that halts the agent's operation) or requires server-side intervention to fix.

💡 Proposed Solution

• The agent should use a version of LangGraph that supports a retry policy, and it should consider implementing this solution when calling the LLM API (also consider implementing this in other parts of the system).
  The agent should return more detailed error messages to give the user a clearer understanding of the underlying issue. This would enable users to better assess whether the problem is temporary or requires further action.
• The agent should return more descriptive error messages to give the user a clearer understanding of the underlying issue. This would enable users to better assess whether the problem is temporary or requires further action.