AI News Summary - March 8, 2026

Anthropic Launches Claude Code Pro with 500K Context Window

Anthropic released Claude Code Pro, expanding the context window to 500,000 tokens specifically for coding tasks. The new tier includes persistent workspace memory that tracks project architecture and dependencies across sessions, reducing repetitive explanations. Early benchmarks show a 40% improvement in multi-file refactoring accuracy compared to the standard Claude Code tier.

MCP Registry Hits 10,000 Published Servers

The Model Context Protocol ecosystem reached a major milestone with over 10,000 community-published MCP servers now available in the official registry. Popular new additions include native integrations for Kubernetes clusters, Salesforce data access, and real-time financial market data. Anthropic and partners announced MCP 2.1 with improved streaming support and bilateral authentication for enterprise deployments.

Cursor Introduces "Team Memory" Feature

Cursor IDE shipped Team Memory, allowing AI coding assistants to learn from collective team coding patterns and style guidelines across shared repositories. The feature uses federated learning to keep code private while building team-specific models. Beta testers report 60% fewer style-related PR comments and faster onboarding for new team members.

Open-Source Agent Framework "Conductor" Reaches 1.0

The Conductor framework for building multi-agent AI systems hit version 1.0 after two years of development. It provides built-in tools for agent orchestration, state management, and human-in-the-loop workflows with support for all major LLM providers. The project has already been adopted by several Fortune 500 companies for internal automation pipelines.