GitHub Transparency Data: Government Takedown Requests and 2026 State Legislative Sessions

Open source in the public sector doesn’t fail because of technical issues. It fails if it’s not aligned with the political agenda https://lnkd.in/dPiWdEsJ

A repository can be labelled open source and still give you nothing to read. One browser agent project is in exactly that state: a single line of description, no README, no license, and no trace of it anywhere else. What caught my attention is how much the description promises and how little it contains. The repository says Scout is Outcrawl's open-source browser agent, mirrored automatically from the company's own repository. That is the whole page. When I tried to open it, both the repository page and the raw README returned a payment-required error. Searches turned up no write-ups, no mirrors, no issue threads. For something labelled open source, that silence says more than the description does. The mirroring detail is the one part worth thinking about. If the public repo syncs from a private one, a bare page is expected early on. That reading is an inference from a single sentence, not a confirmed fact. I would still be careful about treating the label as a description of reality. Plenty of real projects start as a name and a license file, so I would not read a blank page as a verdict either. What evidence would you need before treating a browser agent as installable rather than merely announced? #OpenSource #AIAgents #BrowserAutomation #SoftwareEngineering Read more: https://lnkd.in/d4YipS7V

The OSI has teamed up again with Perforce Zend and OpenLogic, as well as the Eclipse Foundation, for the 2027 State of Open Source Survey.... https://lnkd.in/drctk2p9

“Open source” is one of the most misused phrases on LinkedIn. I keep seeing posts that say “A is now open source” or “an open-source alternative to B.” Some of them are right. A lot of them are describing something else. As a software engineer, I look at open-source alternatives to mainstream software almost every day. Open source has clearly caught everyone’s interest, and that’s exactly why the term gets stretched. Public code is like a book in a library. You can read it, but you can’t photocopy it, rewrite a few chapters and publish your own edition. Open-source code is a book where the author has given you written permission to do exactly that, as long as you follow a few conditions. That written permission is the license. Putting code on GitHub doesn’t make it open source, and neither does making it free to download. If a repo has no license, the default is “all rights reserved,” and reusing the code puts you on shaky legal ground. For software to count as open source, its license has to let anyone: - Use it for any purpose - Study the source code - Modify it - Share it, including modified versions The Open Source Initiative (OSI) maintains the definition most people go by. One rule catches a lot of projects out: the license can’t discriminate. “Free for personal use only” or “anyone except competitors” fails that test. Those projects are usually called source-available. A real example: in 2023, HashiCorp moved Terraform to a license that restricts competing products. The code was still public, but companies building on it suddenly had new rules to follow. The community forked the last openly licensed version into OpenTofu. That’s why it’s worth checking the license before you depend on a tool. A quick check: can you use it commercially, modify it, and share your version? If any answer is no, it isn’t open source. Open-source licenses also differ from each other. MIT, Apache 2.0 and GPL set quite different conditions, and that needs its own post. The full definition is here: https://opensource.org/osd Which open-source tool did you discover recently, and what made you pick it over the mainstream option? #OpenSource #SoftwareEngineering #SoftwareLicensing

I’m building Sin Línea, an AI-assisted news platform focused on Argentina. One question has shaped the project: how can readers inspect the evidence behind what they’re reading? The system brings together news ingestion, research, claim verification, writing, and an audit before publication. I’m also working on the interface that lets readers explore the sources and evidence behind individual claims. One technical decision I found particularly interesting: published articles need a stable record of their supporting evidence. A later evaluation shouldn’t silently change the evidence shown alongside an earlier published version. My day-to-day work focuses on WordPress and React. For this project, I’m using Next.js, FastAPI, PostgreSQL, Celery, and Redis, which has given me a broader set of integration and deployment problems to work through. I’ve now defined the first release’s scope and remaining tasks. My next steps are testing an asynchronous user interaction and checking the reading experience before deployment. When reading a news article, what would you want to see to judge whether a claim is well supported? #WebDevelopment #NextJS #BuildInPublic

A side project I don’t talk about much: OpenSourceDrop. It started as a list I kept while building Sentasity. Every open source tool I tried got a few notes: does it work, what does it really cost, what’s the catch. The catch is usually the interesting part, because open source doesn’t always mean free. Cline, the open source coding agent, costs nothing to install, but every task it runs burns API tokens that you pay for. Licenses, gated features, and paid backends can all change the real price. Today it’s a weekly roundup of new open source tools, the catches to watch for, and open source alternatives to tools you’re paying for. It makes no money and nothing in it is paid. And subscribers NEVER get marketing emails. It’s only ever the tools. Mostly it’s for people like me trying to find good tools, and a small thank you to the people who build them. Link in the comments if you’re curious.

A client needed contact details and reports for 1,300 nonprofits. Done by hand, that is open a website, find the email, find the business number, hunt for the annual report, download it, record everything in a sheet. Then do it 1,299 more times. Weeks of work for one person, and out of date as soon as it is finished. So I built a crawler that collects five things from each organisation in one pass: the name, the CRA business number, a working contact email, the latest annual report, and the latest audited financial statement. First, the target list builds itself. Instead of a manually maintained spreadsheet, the script paginates Imagine Canada's public directory of accredited organisations. The hard part was the emails. Many sites hide them with a Cloudflare trick that replaces the address with a scrambled code. A person sees the email on the page, but a scraper sees nothing useful. I wrote a small decoder that reverses the encoding and recovers the real address. I also added four ways to find an email, used in order: footer links, decoded Cloudflare emails, a scan of the full page, and finally the Contact Us page. If one fails, the next runs. For the PDFs, it matches reports by keyword and filename, always keeps the newest year, and downloads in small pieces with automatic retries. Every result goes straight into a shared Google Sheet. The client opens one sheet instead of 1,300 websites. Case study: https://lnkd.in/dn4M48GU What research does your team still do by hand? #LeadGeneration #Python #WebScraping #Automation

Digital sovereignty is a priority for 94% of organisations in new European open source research, but governance gaps and private software forks carry measurable costs.

I made a tool for migrating Tumblr blogs to Eleventy static sites, since one didn't seem to exist! It retrieves all your posts with all attached metadata, using a Tumblr blog export ZIP and the Tumblr API, and saves them in a new directory structure with Eleventy's data format. This was a quick-and-dirty thing for my own blog's migration, which I decided against for a few reasons, but maybe it will help others. https://lnkd.in/eKus3jJm

Your private JavaScript registry can now publish to the public. Public packages are now GA. Publish and install the packages you want to share with the ecosystem. Read the full announcement here ↓ https://lnkd.in/gvsCd64j