AI News Brief - March 13, 2026

Claude Code Launches Multi-Repository Context Feature

Anthropic has released a major update to Claude Code that allows developers to work across multiple repositories simultaneously while maintaining full context. The new feature uses an enhanced MCP implementation to track dependencies and changes across codebases, reducing context switching by up to 60%. Early adopters report significant productivity gains in microservices architectures.

Open-Source Agent Framework "Conductor" Reaches 1.0

The Conductor framework, which simplifies building autonomous AI agents with tool-use capabilities, has officially hit version 1.0 after 18 months of development. It features built-in support for MCP servers, parallel task execution, and a novel "agent orchestration graph" that lets developers visually design multi-agent workflows. The project has already garnered 15K GitHub stars in its first week of stable release.

GitHub Copilot Workspace Adds Real-Time Collaboration

GitHub announced that Copilot Workspace now supports real-time collaborative coding sessions where multiple developers and AI assistants can work simultaneously on the same codebase. The feature includes conflict resolution powered by GPT-5 and maintains individual context awareness for each participant. Premium tier subscribers gain access starting today.

MCP Consortium Publishes Security Best Practices Guide

The newly formed MCP Consortium, backed by Anthropic, Microsoft, and Google, released comprehensive security guidelines for implementing Model Context Protocol in production environments. The 50-page document covers authentication patterns, sandboxing strategies, and audit logging requirements. Major cloud providers have committed to supporting these standards by Q3 2026.