AI News - August 26, 2026

Claude Code Introduces Multi-Repository Context Sharing

Anthropic announced that Claude Code now supports simultaneous context awareness across up to 10 linked repositories, enabling developers to refactor code across microservices architectures more efficiently. The update includes automatic dependency mapping and cross-repo impact analysis. Early beta testers report 40% faster completion times for distributed system updates.

MCP 2.3 Specification Adds Streaming Data Sources

The Model Context Protocol consortium released version 2.3, introducing native support for real-time data streams including database change feeds, log tails, and event buses. This enables AI agents to maintain continuous awareness of system state changes without polling. Over 30 major AI platforms have committed to implementing the new spec within 60 days.

Open-Source Agent Framework "Nexus" Reaches 1.0

The community-driven Nexus framework achieved its stable 1.0 release, offering production-ready multi-agent orchestration with built-in fault tolerance and rollback capabilities. Nexus supports all major LLM providers and includes pre-built templates for common workflows like data pipeline management and customer service automation. The project has grown to over 15,000 GitHub stars since its alpha launch in May.

GitHub Copilot Workspace Adds Natural Language Testing

GitHub unveiled automated test generation in Copilot Workspace that converts plain English test requirements into full test suites with fixtures, mocks, and edge cases. The feature supports Python, TypeScript, Java, and Go, with measured 85% accuracy on standard test scenarios. Developers can iteratively refine tests through conversational feedback.