AI News Summary - October 16, 2026

Claude Code Introduces Multi-Repository Context Window

Anthropic announced that Claude Code now supports simultaneous analysis across up to 10 linked repositories with a 500K token context window. The feature uses intelligent caching to maintain performance while allowing developers to refactor code across microservices architectures. Early adopters report 40% faster cross-service debugging workflows.

MCP 3.0 Specification Released with Bidirectional Tool Calling

The Model Context Protocol consortium published MCP 3.0, introducing bidirectional tool protocols that allow servers to request capabilities from clients. This enables more sophisticated multi-agent workflows where tools can negotiate permissions and share context dynamically. Over 200 MCP-compatible tools have already committed to supporting the new spec.

OpenHands Agent Framework Reaches 1.0 Milestone

The open-source OpenHands project released version 1.0, featuring stable APIs for autonomous software development agents. The framework now includes built-in safety guardrails, cost management controls, and integration with all major LLM providers. The project has seen 15K GitHub stars since its beta launch in early 2026.

Cursor Launches Team Memory Workspace

Cursor unveiled a shared memory system that allows development teams to build persistent knowledge bases from their codebases and conversations. The system automatically indexes code patterns, team conventions, and architectural decisions, making them available to all team members' AI assistants. Initial benchmarks show 60% reduction in repetitive questions during onboarding.