GitHub MCP Server Adds Pre-Commit Scanning to Bolster AI-Assisted Development Security

Cursor has disclosed the architecture behind Continuity, the system that powers its Origin code-hosting service. The design targets a workload emerging from AI coding agents, which can create large numbers of small repositories far more rapidly than conventional development teams. That pattern puts pressure on traditional Git infrastructure, where storage, metadata and compute are often tightly coupled, making predictable scaling increasingly important as autonomous software development expands.

Continuity uses a write-ahead log stored in Amazon S3 as the authoritative record, while local Git repositories operate as rebuildable caches. Cursor said the separation allows replica counts to rise or fall with usage, decoupling durable storage from serving capacity. The architecture is intended to preserve read and write performance while handling bursts of repository creation, giving Origin a horizontally scalable foundation for agent-driven workloads without treating any individual local Git copy as the final source of truth.

AI coding agents need a secure execution layer to read and modify files, invoke tools and carry out software-development tasks. Vercel built fx in the Zig programming language as a lightweight, portable and model-agnostic environment, aiming to reduce the effort required to add agentic capabilities across command-line workflows, code editors and browser-based applications.

Vercel has now released fx as open-source software. Developers can run it as a standalone command-line tool, embed it in existing development environments or deploy it in a browser through WebAssembly. The project also supports the Model Context Protocol, or MCP, and subagent extensions, allowing teams to connect external tools and assemble multi-agent workflows without committing to a particular AI model.

GitHub Copilot code review has used an agentic architecture running on GitHub Actions since March 5, 2026. It uses a GitHub-hosted runner by default, but can instead use a self-hosted or larger runner. The computing environment for AI reviews, the content they can access and the rules governing teams have therefore become important elements of corporate governance and cybersecurity controls.

GitHub announced an update on June 12, 2026, allowing organization administrators to centrally specify and lock a default runner across all repositories. Copilot will also follow content-exclusion rules at the repository, organization and enterprise levels. GitHub has removed the previous 4,000-character reading limit for custom instruction files in the .github directory and has not announced any additional fees.