GitHub Trending: Today — Immich, Qlib, Trivy, CUDA Samples, Cookbook

immich-app/immich — keeping the photo library local

Immich is for people who want a photo and video library they operate themselves instead of handing it to a consumer cloud. Its large daily increase fits a recurring developer concern: personal media is both data and an application, so a backup-only tool is not enough.

The older HN discussions found by the research script show the practical trade-off. Readers praised local search and context-aware browsing, while others worried that a backup product has to be exceptionally stable and questioned its memory needs. Those are not objections to self-hosting; they are the cost of making it a primary home for irreplaceable files.

aquasecurity/trivy — one scanner across the delivery path

Aqua Security’s Trivy targets teams that need to inspect containers, Kubernetes, repositories, and cloud configuration without swapping tools at every boundary. It checks for vulnerabilities, misconfiguration, secrets, and software bills of materials.

Its community record is unusually sharp because the script surfaced a March security incident and discussion about supply-chain exposure. One HN commenter’s operational advice was to pin GitHub Actions to commit SHAs rather than tags. The current trend entry is not proof of a new incident, but the history explains why a security scanner’s own release and CI practices matter as much as its detection features.

microsoft/qlib — research infrastructure for quant models

Microsoft Qlib is for quantitative researchers who need a platform to move from ideas to production-oriented modeling. The project supports supervised learning, market-dynamics modeling, and reinforcement learning rather than prescribing a single strategy.

The collected HN items are older and have no substantive comment thread, so there is no fresh community verdict to attach to the current bump. Its presence alongside CUDA examples and API recipes is still a useful signal: GitHub’s daily list is rewarding inputs into technical work, not just end-user products.

The useful commonality is ownership of the interface

The five repositories address different jobs, but each gives a developer a handle on a complex system: their own media library, a market-research workflow, a software supply chain, GPU behavior, or an API integration. That is a more grounded theme than calling the list an AI wave. The projects let users inspect, run, and adapt the layer that otherwise stays opaque.