GitHub Trending Daily — 2026-05-28
The daily signal shifted: code understanding holds, but taste/skill packaging and video generation surged into the top ranks.
Snapshot
Today’s top 10 by daily star velocity:
| # | Repo | ★ today | Lang | Category |
|---|---|---|---|---|
| 1 | Lum1104/Understand-Anything | 4,466 | TypeScript | Code knowledge graphs |
| 2 | Leonxlnx/taste-skill | 2,715 | Shell | AI taste guardrails |
| 3 | DigitalPlatDev/FreeDomain | 2,223 | HTML | Free domains |
| 4 | affaan-m/ECC | 2,062 | JavaScript | Agent harness optimization |
| 5 | harry0703/MoneyPrinterTurbo | 1,737 | Python | AI video generation |
| 6 | obra/superpowers | 1,680 | Shell | Agentic skills framework |
| 7 | byoungd/English-level-up-tips | 1,133 | — | English learning guide |
| 8 | mukul975/Anthropic-Cybersecurity-Skills | 885 | Python | Cybersecurity agent skills |
| 9 | Axorax/awesome-free-apps | 728 | JavaScript | Free app curation |
| 10 | anthropics/knowledge-work-plugins | 695 | Python | Claude Cowork plugins |
Trend Interpretation
Understand-Anything continues its multi-day reign at the top, though its star velocity dipped slightly (4,466 vs 4,721 yesterday). It has accumulated a substantial daily-star lead over the rest of the chart, suggesting steady interest in code knowledge graph tooling.
The real story today is the taste/skill layer surging. Taste-skill jumped from 1,440 to 2,715 stars/day and climbed from 7th to 2nd, while Stop-slop remained visible (664 today). The new entry obra/superpowers (1,680/day) — an agentic skills framework — reinforces the same theme. The ecosystem is actively building guardrails for how AI outputs should sound, not just what they should do.
ECC (agent harness optimization) held steady around 2,000/day, confirming yesterday’s observation that the harneess-performance layer has audience stickiness.
MoneyPrinterTurbo (1,737/day) and Kronos (402/day) are the most notable new technical entries — one for AI video generation automation, the other for financial-domain foundation models.
English-level-up-tips (1,133/day) and twenty (520/day) represent the non-agentic tail of the chart, showing general developer tooling (CRM, learning) still has consistent demand.
Dropped from the chart: ai-engineering-from-scratch, jellyfin, OpenStock, claude-mem, ppf-contact-solver. Notably, claude-mem (persistent context agent) fell off after a single day.
Key Signal
The cluster around AI taste/skill packaging (taste-skill, stop-slop, superpowers, ECC) has become the dominant multi-day theme, with 4 of the top 6 repos touching this space. Code knowledge graphs (Understand-Anything) remain the single strongest project, but the surrounding ecosystem is clearly converging on how to control agent output quality through skill infrastructure — not just agent frameworks.
See also: 2026-05-27-github-trending-daily, 2026-05-26-github-trending-daily