Commit graph

8 commits

Author SHA1 Message Date
Jared Liu
4c664cb284 chore: use hi-res og-hq.png for README hero image 2026-04-20 21:09:16 +08:00
github-actions[bot]
9b7b44fbf5 docs: auto-update README [skip ci] 2026-04-20 13:04:56 +00:00
github-actions[bot]
37122fb55e docs: auto-update README [skip ci] 2026-04-20 12:45:07 +00:00
github-actions[bot]
54c7d9ae30 docs: auto-update README [skip ci] 2026-04-20 12:20:28 +00:00
Jared Liu
3bea8cd20f feat: add OG hero image at README top + fix English URL to have no locale prefix
- Embed clickable OG hero image at the very top of each locale README, linking
  to the prompts gallery (respecting the locale — en → no prefix, zh → zh-CN,
  others → their full code, matching youhome's next-intl `as-needed`).
- Introduce `buildPromptsUrl()` helper; consolidate all gallery URL construction
  through it. Fixes the previous English URLs which wrongly emitted `/en-US/`.
2026-04-20 20:19:37 +08:00
github-actions[bot]
02344b8edc docs: auto-update README [skip ci] 2026-04-20 10:21:27 +00:00
Jared Liu
43a896de2c fix: attribute GPT Image 2 to OpenAI (not Google), fix prompt count
- 16 locales: subtitle now correctly reads "OpenAI's GPT Image 2"
  (and proper grammar variants: "de l'OpenAI", "di OpenAI", etc.)
- seedancePromo: bump the nano-banana-pro prompt count from 400+ to
  10000+ (matches the actual awesome-nano-banana-pro-prompts size)
- Regenerated all 16 README_*.md from live CMS

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-20 14:58:05 +08:00
Jared Liu
a00a14c74f feat: scaffold awesome-gpt-image-2 from nano-banana-pro template
Mirror of awesome-nano-banana-pro-prompts tuned for OpenAI's upcoming
GPT Image 2 (codename "duct-tape"):

- Swap CMS model filter to gpt-image-2 + campaign gpt-image-2-prompts
- i18n.ts: all 16 languages rebranded; rewrite seedancePromo to
  cross-promote nano-banana-pro, and whatIs content to reflect
  GPT Image 2's actual strengths:
    * Pixel-perfect multi-language text rendering (zh/en/ja)
    * Cross-image pixel-level consistency
    * Commercial-ready illustration quality
    * True art style induction
  Primary languages (en, zh, zh-TW, ja, ko) hand-translated;
  others fall back to English copy (will be refined later)
- markdown-generator: cover image path switched to
  gpt-image-2-prompts-cover-{en,zh}.png, arenaUrl points at the
  gallery page (the side-by-side arena page doesn't exist yet)
- GitHub Actions:
    * update-readme.yml runs on cron 0 0,12 * * * (twice daily
      per product request, instead of every 4 hours)
    * sync-approved-to-cms.yml wired to gpt-image-2 model
- Cover images, issue templates, docs, LICENSE (year 2026) updated
- Initial 16 README_*.md files generated from live CMS (78 prompts,
  44 categories) so the repo renders immediately without waiting
  for the first Action run

Secrets the repo needs before the Action runs on the remote:
  - CMS_HOST
  - CMS_API_KEY

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-20 14:42:24 +08:00