python: Q regarding support for Prompt Caching in BedrockChatCompletion

[joshuamosesb]

While using BedrockChatCompletion + ChatCompletionAgent (in py API), is Prompt Caching feature of bedrock supported by semantic kernel py-api for the converse and invoke operations?

[joshuamosesb]

one approach could be, in bedrock_chat_completion.py: _prepare_settings_for_request based the self.ai_model_id and whether it supports prompt-caching, all the 4 functions mapped in MESSAGE_CONVERTERS dict could be updated / overloaded (either with optional parameter or to support prompt-caching by default) to include the cachePoint marker

                "cachePoint": {
                    "type": "default"
                }
            }

in their respective return value, to enable prompt-caching in the request .
Pls review and share feedback, or post another comment if there could be any other better approach.