Implement AWS Bedrock Provider via Converse API

[junaidbhura]

Closes #180

Summary

This PR introduces the AwsBedrockProvider and its associated infrastructure. By leveraging the AWS Bedrock Converse API, this implementation provides a unified interface for multiple foundation models (Anthropic Claude, Meta Llama, Mistral, etc.) while adhering to enterprise-level security and compliance standards.

Key Implementation Details

• Authentication: Implements AwsBedrockApiKeyRequestAuthentication. This utilizes Bedrock’s native API Key (Bearer Token) support, ensuring the SDK remains lightweight without requiring the full AWS PHP SDK or SigV4 signing.
• Unified Converse API: Maps the library's internal structures to Bedrock's converse and converse-stream logic.
• Multimodal Support: Full support for image and document inputs via MessagePart and File DTOs.
• Tool Calling: Reliable mapping of FunctionDeclaration to Bedrock's toolConfig, including support for toolResult and toolUse blocks.
• Dynamic Discovery: The AwsBedrockModelMetadataDirectory automatically infers model capabilities (Text, Image, Tools) from the AWS Control Plane.

Technical Architecture

• Dual-Plane URLs: Correctly separates Control Plane (bedrock.region.amazonaws.com) for model listing and Runtime (bedrock-runtime.region.amazonaws.com) for inference.
• Finish Reason Mapping: Bedrock-specific stopReason values are mapped to the library's FinishReasonEnum.

---

Testing & Verification

I have verified this implementation against the following scenarios using the anthropic.claude-3-5-sonnet-20241022-v2:0 model:

Test Case	Status	Notes
Simple Text Generation	✅ Pass	System instructions and temperature respected.
Token Usage Tracking	✅ Pass	Metadata correctly extracted from usage object.
Stop Sequences	✅ Pass	Model successfully halts at custom stop tokens.
Multimodal (Vision)	✅ Pass	Successfully described PNG/JPEG images via Base64.
Tool Calling	✅ Pass	Correctly parsed toolUse arguments into FunctionCall DTOs.

Example Usage

// Set `AWS_BEDROCK_API_KEY` and `AWS_BEDROCK_REGION` environment variables 

use WordPress\AiClient\AiClient;
use WordPress\AiClient\ProviderImplementations\AwsBedrock\AwsBedrockProvider;

// The provider resolves the region via AWS_BEDROCK_REGION env variable
$result = AiClient::prompt('Summarize the benefits of using a unified AI SDK.')
    ->usingModel(AwsBedrockProvider::model('anthropic.claude-3-5-sonnet-20241022-v2:0'))
    ->usingTemperature(0.7)
    ->generateTextResult();

echo "Tokens Used: " . $result->getTokenUsage()->getTotalTokens();

[github-actions]

The following accounts have interacted with this PR and/or linked issues. I will continue to update these lists as activity occurs. You can also manually ask me to refresh this list by adding the props-bot label.

Unlinked Accounts

The following contributors have not linked their GitHub and WordPress.org accounts: @junaidbhura.

Contributors, please read how to link your accounts to ensure your work is properly credited in WordPress releases.

If you're merging code through a pull request on GitHub, copy and paste the following into the bottom of the merge commit message.

Unlinked contributors: junaidbhura.

To understand the WordPress project's expectations around crediting contributors, please review the Contributor Attribution page in the Core Handbook.