AI News Summary – April 10, 2026

Claude Code Launches Multi-Repository Context Feature

Anthropic released an update to Claude Code enabling simultaneous context awareness across up to 10 linked repositories. The feature uses advanced caching to maintain coherent understanding of complex microservice architectures while reducing token costs by 40%. Early adopters report significant improvements in refactoring tasks that span multiple codebases.

MCP 2.0 Specification Enters Public Review

The Model Context Protocol consortium published the draft MCP 2.0 specification, introducing bidirectional streaming and enhanced security primitives for AI agent communication. The update addresses previous limitations in real-time collaboration scenarios and adds native support for multi-modal context sharing. Major implementations from Anthropic, Microsoft, and Google are expected within 60 days.

Open-Source Agent Framework "Nexus" Reaches 50K GitHub Stars

Nexus, a lightweight Python framework for building autonomous AI agents, hit a major adoption milestone after just three months. The framework's appeal lies in its minimal boilerplate code and native MCP integration, allowing developers to create production-ready agents in under 100 lines. Contributors recently added support for hierarchical agent teams and improved error recovery mechanisms.

Cursor Introduces AI-Powered Debugging Copilot

Cursor unveiled "Debug Mode," an AI assistant specifically trained on millions of debugging sessions to identify and fix runtime errors. The feature integrates with popular observability platforms to analyze stack traces, logs, and application state before suggesting fixes. Beta testers report 60% faster resolution times for complex production issues.