AI News - March 22, 2026

Claude Code Introduces Multi-Repository Context Windows

Anthropic announced Claude Code now supports simultaneous context awareness across up to 10 linked repositories, with a combined 500K token context window. The feature enables developers to refactor code across microservices architectures while maintaining consistency. Early adopters report 40% fewer context-switching errors in distributed system development.

MCP Protocol Reaches 1.0 Stable Release

The Model Context Protocol consortium released version 1.0, marking the first production-ready standard for AI tool integration. Over 200 development tools now support MCP natively, including major IDEs, documentation systems, and CI/CD platforms. The standardization is expected to significantly reduce integration complexity for enterprise AI deployments.

Open-Source Agent Framework "Cortex" Gains Traction

The newly released Cortex framework has surpassed 15K GitHub stars in its first week, offering a lightweight alternative to LangChain and AutoGPT. Built with TypeScript-first design, it provides native streaming support and built-in observability for multi-agent systems. Contributors highlight its 3x faster initialization times compared to existing frameworks.

GitHub Copilot Workspace Adds Automated PR Reviews

GitHub expanded Copilot Workspace with AI-powered pull request reviews that automatically check for security vulnerabilities, coding standards, and test coverage gaps. The feature uses a fine-tuned model trained on 50M+ reviewed PRs and integrates directly with GitHub Actions. Beta users report reducing review cycles by an average of 2 days.