MCP Project Orchestrator

# AWS Integration This document covers AWS integration capabilities for the MCP Project Orchestrator. ## AWS MCP Integration The MCP Project Orchestrator includes comprehensive AWS MCP (Model Context Protocol) integration that enables AI-powered access to AWS services. For detailed information, see [AWS_MCP.md](./AWS_MCP.md). ### Quick Start 1. Install AWS dependencies: ```bash pip install -e ".[aws]" ``` 2. Configure AWS credentials: ```bash export AWS_REGION=us-east-1 export AWS_ACCESS_KEY_ID=your_key export AWS_SECRET_ACCESS_KEY=your_secret ``` 3. Start the MCP server with AWS integration enabled. ### Features - **AWS Service Access**: S3, EC2, Lambda, CloudFormation, IAM - **Best Practices**: Security, cost optimization, performance guidelines - **Cost Estimation**: Predict AWS costs based on usage - **Documentation**: Contextual AWS guidance and examples See [AWS_MCP.md](./AWS_MCP.md) for complete documentation. ## AWS Artifacts and Container Publishing This project uses GitHub OIDC to assume an AWS role for publishing artifacts. ### Required AWS setup - Create S3 bucket: `mcp-orchestrator-artifacts` - Create ECR repository: `mcp-project-orchestrator` - Create an IAM role with trust policy for GitHub OIDC and permissions to push to ECR and write to S3. Save its ARN as GitHub secret `AWS_OIDC_ROLE_ARN`. ### Release workflow On GitHub release publish: - Build and push container image to ECR using Podman - Build Python dists and upload to S3 under `releases/<tag>/` - Optionally archive Conan cache to S3 See `.github/workflows/release.yml`.