
GitHub has announced the general availability of its secret scanning feature within the MCP server, aimed at enhancing code security for developers. This tool allows users of AI coding agents like GitHub Copilot CLI and Visual Studio Code to scan for exposed secrets before committing code, thus preventing credential leaks. The feature, previously in public preview since March 2026, respects existing push protection settings, ensuring consistent security measures. This development underscores GitHub's commitment to integrating AI tools for improved code security.
Read original
© GitHub ChangelogGitHub has introduced a public preview for enterprise-managed plugins in its Copilot CLI, allowing administrators to configure and distribute plugins across their organizations. This move aims to streamline developer onboarding and reduce setup time by enabling automatic installation of custom agents and skills. Enterprises can now set baseline standards and governance strategies through hooks and MCP configurations, ensuring consistent plugin use across all users. This update enhances the flexibility and control enterprises have over their development environments, making it easier to manage and deploy custom tools at scale.
© GitHub ChangelogGitHub Copilot's latest updates for Visual Studio Code bring a suite of enhancements aimed at improving developer productivity. Notably, Copilot now supports semantic search across any workspace and allows grep-style queries across GitHub repositories. The introduction of the experimental /chronicle feature enables users to query their chat history, providing a personalized workflow experience. Additionally, the updates include smarter prompt caching and deferred tool loading to reduce token usage, while new agent capabilities like inline diffs in chat and browser tab sharing enhance collaboration. These changes make Copilot a more integrated and efficient tool for developers working in VS Code.
© GitHub ChangelogGitHub's MCP Server now includes a public preview of its dependency scanning feature, allowing developers to catch vulnerabilities before committing code. This tool integrates with MCP-compatible IDEs and AI coding agents, providing real-time alerts on known vulnerabilities. By leveraging the GitHub Advisory Database, it offers structured results with severity levels and recommended fixes. This development enhances the security workflow for developers, making it easier to address potential issues early in the coding process.
Claude Code's latest update, v2.1.129, introduces several enhancements and fixes that improve user experience and functionality. Notably, a new flag allows fetching plugins directly from a URL, and an environment variable ensures synchronized output on certain terminals. The update also refines plugin manifest declarations and adjusts model discovery settings to be opt-in. These changes, alongside numerous bug fixes, make the platform more robust and user-friendly, particularly for developers using third-party deployments or managing complex projects.
The latest update to Claude Code, version 2.1.128, introduces several enhancements and bug fixes aimed at improving user experience and functionality. Notably, the update allows the plugin directory to accept .zip archives, enhancing flexibility in plugin management. Additionally, subprocesses no longer inherit OTEL environment variables, which streamlines the operation of OTEL-instrumented apps. The update also addresses various bugs, such as fixing issues with image uploads and terminal notifications, ensuring smoother and more reliable performance. These changes make Claude Code more robust and user-friendly, particularly for developers relying on its tools.
The latest vLLM release replaces the deadsnakes PPA with building Python from source to improve performance.