When the AI provider returns an error (e.g. HTTP 529 overloaded, 429 rate limit), OpenClaw surfaces a generic message like: Include the provider name in the built-in error messages. For example:

• "OpenAI API error (500). Please try again." In the error handling layer that generates user-facing messages, append the provider name from the active model config before sending to the messaging surface.

TL;DR

Modify the error handling layer to include the provider name in the error messages sent to users.

Guidance

• Identify the error handling layer responsible for generating user-facing messages and locate where the provider name is accessible.
• Append the provider name from the active model config to the error message before sending it to the user.
• Consider using a template or formatting function to consistently include the provider name and error code in the message.
• Review the code to ensure that the provider name is correctly retrieved and appended for each supported provider (e.g., Anthropic API, Groq, OpenAI).

Example

# Assuming a function generate_error_message(error_code, provider_name)
def generate_error_message(error_code, provider_name):
    error_messages = {
        529: f"{provider_name} is temporarily overloaded. Please try again in a moment.",
        429: f"{provider_name} API rate limit reached. Please try again shortly.",
        500: f"{provider_name} API error. Please try again."
    }
    return error_messages.get(error_code, "Unknown error")

Notes

This solution assumes that the provider name is available in the active model config and can be accessed in the error handling layer. The exact implementation may vary depending on the programming language and framework used.

Recommendation

Apply the suggested change to the error handling layer to provide more informative error messages to users, allowing them to better diagnose and respond to issues with specific providers.