[FEATURE] Cache Strategy

[Unshure]

Problem Statement

As a Strands user, I would like to be able to apply a caching strategy to my agent or model provider so that it applies some caching strategy to each request.

Proposed Solution

Add a cache_config parameter to the BedrockModelProvider, AnthropicModelProvider, and any other model provider that applies a caching strategy to the messages array.

Use Case

• Apply some default cache strategy to help reduce token usage in bedrock or other model providers

Alternatives Solutions

• Build in a caching strategy into the agent class

Additional Context

• https://docs.aws.amazon.com/bedrock/latest/userguide/prompt-caching.html
• https://docs.anthropic.com/en/docs/build-with-claude/prompt-caching
• https://platform.openai.com/docs/guides/prompt-caching

[Unshure]

First step of this cache strategy feature has been merged into the bedrock model provider: #1438

Follow up work includes:

• Adding support for other models, like nova, into the bedrock model provider: [FEATURE] Create Nova model cache strategy #1571
• Add cache strategy to other model providers (LiteLLM, Anthropic, OpenAI, etc)

[Unshure]

Initial docs for cache strategy: strands-agents/docs#484