What’s New on GitHub Today – 14 Nov 2026

Here’s today’s look at what developers are starring on GitHub, plus the latest product updates from the GitHub changelog. Star counts below are exactly as shown on GitHub Trending and its daily language pages (Python, JavaScript, TypeScript, PHP) on the evening of 14 November 2026 (IST). They change every hour, so expect different numbers when you click through.

The big theme: agents that work like coworkers, with guardrails

Agent tooling still dominates Trending, but today’s picks lean towards running agents responsibly: giving them real browser debugging tools, putting human approval gates in the loop and red-teaming the AI systems you ship. Alongside them, some reliable self-hosted classics (uptime monitoring and machine translation) show that practical infrastructure still earns stars.

Trending repositories

1. ChromeDevTools/chrome-devtools-mcp

“Chrome DevTools for coding agents.” This official MCP server from the Chrome DevTools team lets an AI coding agent control and inspect a live Chrome browser, the same way you would use DevTools yourself.

• Language: TypeScript
• Stars: 56,082 (673 today)
• Why devs care: agents that can only read source code guess at front-end bugs. Giving them real DevTools access to the console, network and performance data lets them check their fixes in an actual browser.

2. andrewyng/openworker

OpenWorker has two parts: a desktop app for everyday work (research, writing, code review and tasks across your existing tools) and a Harness SDK for building your own agentic desktop and web apps. According to its README, you bring a cloud model key or run locally with Ollama, set permissions and choose which actions need your approval. The desktop app is in open beta.

• Language: Python
• Stars: 19,178 (94 today)
• Why devs care: it pairs a ready-to-use “AI coworker” with an SDK, and it puts approvals and permissions front and centre, which matters as soon as an agent can touch real files and accounts.

3. dream-num/univer

The Univer Office SDK lets you “build your own AI office”: it brings editing, calculation, collaboration and file processing into your product for both people and AI agents, on an extensible open-source core.

• Language: TypeScript
• Stars: 24,306 (495 today)
• Why devs care: embedding spreadsheet and document editing in a web app is hard to build from scratch. An open-source SDK designed with agent access in mind is a strong base for internal tools and SaaS dashboards.

4. Tencent/LoopForge

Self-hosted workflows for delivering code with coding agents. LoopForge orchestrates CodeBuddy, Codex and Claude Code with scripts and human approval gates, and includes local workflows for coding assistants.

• Language: Go
• Stars: 760 (174 today)
• Why devs care: most teams want agents to do the busywork but still want a person to sign off before code ships. A self-hosted orchestrator with explicit approval steps is a sensible pattern to study, even at this early stage.

5. louislam/uptime-kuma

“A fancy self-hosted monitoring tool.” Uptime Kuma checks your websites and services and alerts you when they go down, from a clean web dashboard.

• Language: JavaScript
• Stars: 94,112 (432 today)
• Why devs care: it’s one of the easiest ways to get uptime monitoring and status pages for side projects, small business sites and homelabs without paying for a SaaS plan.

6. metadist/synaplan

An open-source AI platform for your knowledge, models and channels: web, WhatsApp, email, Outlook, Nextcloud and ownCloud. It runs local and cloud models side by side and shows the cost on every answer.

• Language: PHP
• Stars: 905 (90 today)
• Why devs care: a rare PHP project in the AI-platform space. Per-answer cost tracking and WhatsApp or email channels are exactly what small businesses ask for when they want an assistant on their own data.

7. LibreTranslate/LibreTranslate

A free and open-source machine translation API: self-hosted, capable of working offline and easy to set up.

• Language: Python
• Stars: 17,703 (130 today)
• Why devs care: if you need translation without sending user text to a third-party API, for privacy, cost or offline use, LibreTranslate gives you a simple HTTP API you can run on your own server.

8. microsoft/PyRIT

Microsoft’s open-source AI red-teaming toolkit. It tests generative AI systems with automated scans, adaptive multi-turn attacks and a web UI.

• Language: Python
• Stars: 4,829 (22 today)
• Why devs care: if you ship a chatbot or agent, someone will try to jailbreak it. Running automated attacks before launch is far cheaper than finding out from users.

GitHub product updates

There are no new changelog entries dated 14 November yet, so here are the most recent updates from 13 November.

Gemini 4 Hydrogen in GitHub Copilot (13 Nov)

Gemini 4 Hydrogen, which GitHub describes as Google’s latest Flash model, is now generally available in Copilot for Pro, Pro+, Max, Business and Enterprise users. It’s aimed at fast agentic coding, tool use and multistep workflows. It’s billed at provider list pricing under usage-based billing and is rolling out gradually across VS Code, Visual Studio, Copilot CLI, the coding agent, JetBrains IDEs, Xcode and more. Business and Enterprise admins control access through the model policy. (changelog)

GitHub Enterprise Server support in the Copilot app (13 Nov)

You can now add a GitHub Enterprise Server account to the GitHub Copilot app and work with its repositories, issues and pull requests. The feature was previously an opt-in preview. It needs Copilot app 1.2.0 or later and a GitHub.com or GHE.com sign-in first. Go to Settings > Accounts > Add account. Cloud sessions aren’t available for GHES. (changelog)

Copilot model deprecations on 14 December (13 Nov)

GitHub will deprecate several models across all Copilot experiences on 14 December 2026: Claude Fable 5 (use Claude Fable 5.2), Claude Haiku 4.5 (use Claude Haiku 5.5; it stays available to individual subscribers on annual plans), Gemini 3.8 Flash (use Gemini 4 Hydrogen), Gemini 4 Argon (use Gemini 4 Carbon), Grok 4.6 (use Grok 4.7) and Kimi K3 (alternative to be announced). Update any workflows or integrations that name these models. (changelog)

How to use this list

• Building front ends with an AI assistant? Try chrome-devtools-mcp so your agent can check its work in a real browser.
• Letting agents touch real code or accounts? Study the approval models in openworker and LoopForge, and red-team with PyRIT before launch.
• Running small sites or a homelab? Uptime Kuma and LibreTranslate are quick, self-hosted wins.
• Using Copilot with pinned models? Move off the deprecated models before 14 December.

Takeaway

Today’s Trending page is about making agents trustworthy, with real debugging tools, human sign-off and adversarial testing, alongside a few dependable self-hosted tools. Star the ones that fit your stack, read the READMEs critically and try one on a side project first. See you in tomorrow’s roundup.