Thanks for the answers.
On the VCS point, I want to flag this for the roadmap rather than let it pass. The Zig project moved off GitHub to Codeberg late last year, explicitly because GitHub Actions became unreliable for their CI pipeline. The Free Software Foundation has an active campaign encouraging open source projects to follow. These are not edge cases. A platform that treats GitHub as the only source of truth will have blind spots for developers who have already made that move, and more will.
On premium extensions: good to hear that is on the horizon. When the time comes, happy to work toward getting Floxum extensions listed alongside the open source ones.
Why GitHub is losing FOSS trust
The slide has been accumulating across several fronts:
- Uptime collapse — 94 incidents in 2023, 119 in 2024, 257 between May 2025 and April 2026 with 48 classified as major. GitHub Actions was the worst-hit service (57 outages in that period). GitHub's own CTO admitted the platform was not ready for AI-driven traffic growth and needed 30x more capacity.
Uptime dropped below 90% in 2025. The Zig project migrated to Codeberg, explicitly citing Actions hanging build servers for months with no fix.
- Copilot trained on your code — In March 2026 GitHub changed policy to default to using Copilot interaction data for AI training. The community response: 59 thumbs-down, 3 rocket ships. A lawsuit over training data from open-source repositories (filed 2022) has survived partial dismissal.
- Microsoft ownership anxieties — Growing concern over vendor lock-in and digital sovereignty, noted explicitly in the 2026 State of Open Source Report. Microsoft's acquisition created a slow but steady perception shift: GitHub went from a community-owned neutral platform to a commercial product with AI monetisation goals.
- AI "slop" flooding repositories — AI-generated issues and PRs have flooded open-source projects. GitHub itself calls it a "DoS attack on human attention." The number of maintainers has not kept pace, and GitHub's own tooling (Copilot) is partly responsible.
- Code quality pressure — Research shows Copilot correlates with more "mistake code" being pushed and less code reuse.