GitHub

More of every Copilot session goes toward useful work, helping your credits go further without changing how you work. In Copilot for Visual Studio Code, prompt caching reuses model state across turns, while tool search loads definitions only when needed. Together, they reduce unnecessary context while keeping a broad toolset available. Auto selects a model using task intent and real-time model health. In a matched three-model evaluation, its HyDRA router matched OpenRouter Auto’s 70.8% resolution rate at 3.3x the savings. Auto also routes at natural cache boundaries, allowing cached context to keep building. Learn how GitHub Copilot is making more of each session. https://lnkd.in/eK3U3d4J

📣 Anthropic's Claude Opus 5 is now available and rolling out in GitHub Copilot. Early testing shows ➡️ It has strong performance on agentic coding workflows ➡️ It's effective at making targeted changes, validating its work, and reducing unnecessary execution overhead on complex tasks Try it out in the GitHub Copilot app or Visual Studio Code 👇 https://lnkd.in/gA-G5FQJ

Age assurance laws are reaching down to operating systems and app stores. Anne Dickison (FreeBSD Foundation), Katie Steen-James (OSI), and Margaret Tucker (GitHub) break down what's happening across the US, Brazil, and Europe, and how maintainers can help shape policy that fits how open source actually works.

Cowork, comment, and chill with the GitHub team.

Over the past three months, three million secrets have been exposed. Many teams don’t realize it until the damage is done. Secret leaks are a growing risk in modern software development, yet many organizations still lack the visibility needed to understand their exposure. Join GitHub Advanced Security for a webinar on July 28 to learn: • Key secret leak trends based on our analysis of GitHub data and practical next steps • How public monitoring can detect exposed secrets across GitHub, not only in your organization’s repositories • How to assess your organization’s secret exposure and take immediate action Register now ⬇️ https://lnkd.in/eFCFApxJ

In this stream we'll experiment with building a personal assistant with GitHub CLI. We'll use workflows and experiment with features

AI made code cheaper to write. Understanding, reviewing, and owning it? Not so much. That changes how engineering teams should evaluate small feature requests. Instead of debating scope based on assumptions, ask for a constrained first patch. Treat it like a price check, not the finished product. AI shouldn’t make the decision. It should help humans gather concrete evidence and make a faster, better-informed one. The cost of saying yes has changed. The cost of saying no should change with it. https://lnkd.in/e5xMM33K

Many teams can tell you whether developers are using AI. Fewer can show how they’re using it and whether those behaviors are associated with measurable engineering outcomes. Drawing on findings from thousands of organizations, we explore adoption trends across surfaces and how progression from code completion to agentic workflows is associated with higher throughput and faster time-to-merge. With the new GitHub Copilot impact dashboard, enterprise leaders can measure those same adoption patterns and engineering outcomes within their own organizations—helping them identify opportunities for enablement, guide investments, and advance adoption. Learn how your organization can connect Copilot adoption to impact and scale with confidence. https://lnkd.in/e_eCxENK

Join us for Open Source Friday as we sit down with Brady Gaster (@bradygaster) to explore Squad, an open source project for building AI agent teams for any project. Repo: https://lnkd.in/e69S4pzH #OpenSourceFriday #GitHub #AI #Copilot #OpenSource