AI News Summary - April 7, 2026

Claude Code Introduces Multi-Repository Context Awareness

Anthropic rolled out a major update to Claude Code that enables simultaneous context understanding across multiple linked repositories. The feature uses an advanced dependency graph to track cross-repo relationships, making it particularly useful for microservices architectures. Early adopters report 40% faster implementation times for features spanning multiple codebases.

MCP 2.0 Specification Released with Bidirectional Tool Calling

The Model Context Protocol consortium published version 2.0, introducing bidirectional tool execution that allows context servers to request actions from connected AI models. This architectural shift enables more sophisticated agent workflows where external systems can proactively trigger model reasoning. Over 200 MCP servers have already committed to supporting the new spec.

OpenHands Agent Framework Reaches Production Readiness

The open-source OpenHands project announced its 1.0 release after 18 months of development, offering a production-grade framework for building autonomous coding agents. The framework includes built-in safety controls, rollback mechanisms, and integration with all major coding assistants. GitHub reports that 15,000 repositories are already using OpenHands for automated maintenance tasks.

Cursor Launches "Shadow Mode" for Silent Code Learning

Cursor introduced an opt-in feature that observes developers' coding patterns without making suggestions, building personalized models of coding style and architecture preferences. After a two-week learning period, the assistant can generate code that matches individual or team conventions with 95% accuracy. The feature has sparked privacy discussions despite being fully local and encrypted.