visual-review
/visual-review
Określa standardy strony z artefaktem oceniane wzrokiem: panele przed i po, nakładki, przełącznik migania i notatkę o sposobie czytania.
- Repozytorium
- claude-config ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 09.11.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 1
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 0
Brak sygnałów. Repozytorium zmienione 09.11.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
Co robi i kiedy
Określa standardy strony z artefaktem oceniane wzrokiem: panele przed i po, nakładki, przełącznik migania i notatkę o sposobie czytania.
Oryginalny opis (EN)
Standards for the artifact page that ends work judged by eye (images, fits, meshes, maps, plots): before/after panels, overlays, blink toggle, subject selection and a how-to-read note. Use when building or reviewing such a page, and when checking renders, plots or screenshots during the work.
Jak działa
- Checking renders during the work
Podgląd SKILL.md
--- name: visual-review description: >- Standards for the artifact page that ends work judged by eye (images, fits, meshes, maps, plots): before/after panels, overlays, blink toggle, subject selection and a how-to-read note. Use when building or reviewing such a page, and when checking renders, plots or screenshots during the work. --- # Visual review page Publish the page as an artifact, with metrics beside the views, never instead of them. - Before and after on the same subjects, same crop, same scale, with an overlay that makes the question checkable (the reference outline over the image) and a blink toggle for A against B. - Subjects chosen by a rule fixed before fitting, spread across the cases, never by how good they look. Worst cases shown beside typical ones. - Spread across places or conditions (one per compass direction, say), with sites that failed listed as failed. - A short "how to read this" at the top: what each panel is, the one question to answer, known artefacts. ## Checking renders during the work A `research-mid` subagent reads the images, not the main thread: an image read in the main thread is resent every turn after. Its brief names the image paths and the criterion each is judged against, and asks for a per-image verdict (pass/fail, what is wrong, where in the image). The main thread reads an image itself only when a verdict is ambiguous or contradicts the metrics.
Wersje
| Repozytorium | Wynik | Licencja |
|---|---|---|
| Ta strona · ChristopherMichael-Stokes/claude-config | 100 | MIT |
| xoai/sage | 100 | MIT |
| colinhacks/frizz | 100 | MIT |
| helderberto/agent-skills | 100 | MIT |
| gilrotem/web-studio | 100 | MIT |
| lucasfcosta/backpressured | 100 | MIT |
Migawka bazy z 13.11.2026 · wersja paczki bd520e15fd18 (skrót SHA-256 całej paczki)
Pobrania w ostatnich 30 dniach: 0 · ranking
Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.
Podobne skille
Brak podobnych skilli w bazie.