4 MIN READ
Every year takes a few tools with it. 2025 was longer than most. Heroku killed free tier dynos. Stoplight sunsetted Spectral v5 in favor of a hosted rewrite the maintainers will not ship the source for. Curl almost lost a maintainer, and the maintainer walked back the resignation a week later, but the week served as a reminder. The small vendor with the niche CLI got absorbed by the larger vendor with the customer list. The open source project with the single maintainer ended when the maintainer burned out. The tool people depended on for a decade got a deprecated notice, a migration guide nobody wrote, and a quiet exit discovered three months too late.
What follows runs as the working version of the obituary, shorter than the version the maintainer would want and longer than the version anyone has time to read. 2025 took the casual Heroku demo, the Spectral CI pipeline, a few monitoring tools the industry had quietly stopped paying for, and one incident response framework that had been on life support since the lead maintainer moved to a startup with no time for the project. The pattern underneath the names stays the same pattern, and the pattern amounts to the lesson.
What disappeared and why
Acquired and absorbed amounts to the first pattern. The small vendor with the niche product gets bought by the larger vendor that wanted the customer list, the product gets absorbed into the larger product, and the workflow built around the small tool breaks in the integration. Stoplight plus SmartBear sits as the 2025 example, Spectral v5 plus the hosted rewrite as the next one. The acquisition amounts to a business event. The migration amounts to a working developer problem, and the migration amounts to what the working developer was not told about in the announcement.
Funding cut runs as the second pattern. The open source project that depended on company sponsorship loses the sponsorship when the company runs out of runway, the project that had three full time maintainers ends up with one part time maintainer, and the project the industry depended on ends up depending on a maintainer’s free evenings. Tidelift’s 2024 layoffs and the 2025 contraction in open source funding serve as the macro version. The micro version amounts to the maintainer of a tooling library announcing on a Thursday that the library now sits in maintenance mode because the company that funded it pivoted.
Maintainer burnout rounds out the three. The single maintainer carries a project for a decade, the project grows, the user base grows, the issue queue grows, and the maintainer’s free time does not. The maintainer burns out. The project ends with a blog post. The industry that depended on the project discovers the dependency the same week. The funding model that would have prevented it amounts to a funding model the industry has been talking about for a decade and not building.
What replaced them
Direct successor amounts to the replacement the maintainers hope for. The new project picks up the user base, the documentation, the maintainers, the migration script runs cleanly, and the working developer moves over without rebuilding the workflow. Spectral v6 serves as the attempt. The migration from v5 to v6 amounts to a documented one, and the maintainers have been clear about the version they will support and the version they will not. The pattern works when the maintainers plan the transition, and the maintainers almost never plan the transition.
Adjacent tool runs as the replacement people actually get. The tool already on the system adds the capability, the migration becomes a config change, and the working developer does not need a new project to do the job. ESLint folded several of the rules the smaller linters used to ship. Terraform absorbed several of the smaller infrastructure tooling libraries. The adjacent tool amounts to the replacement that does not generate a blog post, which amounts to the reason it does not get counted as a replacement when the tally comes out.
Role elimination sits as the quietest replacement. The workflow the old tool supported ended because the workflow itself became obsolete. The Kubernetes manifest editor serves as the 2025 example. Helm, Kustomize, and the IDE-native tooling absorbed most of the standalone editor use case, and the standalone editor no longer amounts to a project anyone needs to track. The tool did not get replaced. The work the tool did got replaced, and the tool went with it.
What the list teaches
Bus factor runs as the first lesson. The project that depended on a single maintainer ended when the single maintainer ended, and the contributor agreement never addressed the risk. The check amounts to a simple one. Look at the commit history for the last two years. Count the maintainers who have committed in that window. One amounts to a risk. Two amounts to a watch. Three amounts to a foundation. The check takes ten minutes and prevents the next obituary.
Vendor lock in runs as the second lesson. The tool chosen because it was the best at the time turned into a tool that could not be left when the tool got acquired, and the open source license was supposed to prevent the lock in that the open source license did not prevent. The check amounts to the contract and the export. The data the tool generates should be exportable in a documented format. The configuration should sit in a text file the working developer owns. The integration should be replaceable without a migration project. The tool that fails any of those checks amounts to a lock in the working developer has not noticed yet.
Maintenance funding rounds out the three lessons. The open source project that gets used every day sits funded by a maintainer’s free time, and the funding amounts to the funding anyone can contribute to. GitHub Sponsors, Open Collective, Tidelift, the funding mechanism does not serve as the point. The point amounts to the funding, full stop. The next maintainer who burns out serves as the next maintainer the industry could have supported. The check amounts to the next use. The next use amounts to the moment to fund it.
The bottom line
Check the bus factor. Test the export story. Fund the maintainer. The 2026 list amounts to a draft already in motion, and the stack that survives it runs as the stack whose owner ran those three checks before the announcement email landed.
Sources & Further Reading
All claims in this article are sourced from primary documentation, vendor advisories, and reputable security researchers.
Spotted an error? Email the editor. Corrections are issued with a visible correction note.
Editorial standards. Every article on humanrequired.org is reviewed by a human editor before publication. AI may assist with drafting or research; final editorial control is human. Read the full standards.
