AI News Brief - October 10, 2026

Claude Code Introduces Multi-Repository Context Windows

Anthropic announced that Claude Code now supports simultaneous context from up to 10 linked repositories, effectively allowing developers to work across microservices architectures without manual context switching. The update includes intelligent dependency mapping that automatically identifies relevant code across repos when answering queries. Early adopters report 40% faster debugging times for distributed systems.

MCP 2.3 Specification Adds Streaming Capabilities

The Model Context Protocol consortium released version 2.3, introducing bidirectional streaming for real-time data synchronization between AI models and external tools. This enables agents to receive live updates from databases, APIs, and file systems without polling, significantly reducing latency in production environments. Major MCP server implementations including Axiom and ContextHub have already shipped compatible updates.

OpenHands Agent Framework Reaches 1.0 Milestone

The open-source OpenHands project released version 1.0 with production-ready multi-agent orchestration and a plugin ecosystem of over 200 community-built tools. Key features include declarative agent workflows, built-in observability with OpenTelemetry integration, and native MCP support for tool connectivity. The framework has been adopted by over 3,000 organizations since its beta launch.

GitHub Copilot Workspace Adds Visual Debugging

GitHub shipped visual debugging capabilities to Copilot Workspace, allowing developers to step through AI-generated code with inline explanations for each logic branch. The feature uses execution traces to highlight which parts of generated code are actually being used and suggests optimizations for dead code paths. The update also includes improved test generation that achieves 85% branch coverage on average.