Adonis0123/adonis-skills / skills/prompt-studio
prompt-studio
FlaggedTurn a one-line image or video request into a finished asset: route it to an image, a code-rendered Code Video or a Model Video, search many GitHub prompt libraries for proven image prompts, video-model prompts (Seedance, Kling, Veo, Hailuo/MiniMax, Wan, Grok Imagine) and code-video templates, show the best 5 with their samples, model, author and license, write a generation brief, hand it to an available channel (host tool, Codex/Grok CLI, API, Hyperframes), then check and log the result. Use whenever the user wants a cover, illustration, poster, thumbnail, infographic, animation, motion clip, AI video, or a better prompt for one, even if they only say 'make me an image'. Also for Chinese-language requests for WeChat covers, illustrations, still-to-motion clips, and image/video prompt optimization. Not for general chatbot or coding prompts, photo editing without generation, UI implementation, or slide decks.
references/sources.md
View on GitHub# Prompt sources
`recommend` searches many GitHub prompt libraries at once, through one index. Each source has an adapter, a license and a seed policy in `assets/sources.json`; `studio.mjs sources [--kind video]` lists them with what is cached. Libraries were checked on 2026-10-29; counts change daily.
## What is searched
| Kind | Sources | Sample media |
| ------------ | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------- |
| `image` | YouMind ai-image-prompts (by use category), YouMind GPT Image 2 (18k), freestylefly/awesome-gpt-image-2, ai-boost (GPT Image, Nano Banana), songguoxs (GPT-4o / Nano Banana), glidea (Nano Banana, Chinese), yang0/handraw-style (367 hand-drawn styles; style traits, not full prompts) | Sample images from the model named on the entry |
| `video` | LearnPrompt Seedance (Seedance 2.x), SkyNotSilent MiniMax H3, ai-boost (Seedance, Veo, Kling, H3, Wan, Sora), imagineVid Kling 3.0 and Grok Imagine, YouMind Seedance, LichAmnesia ad prompts | Playable clip and poster on most entries |
| `code-video` | Hyperframes catalog (templates, `npx hyperframes add <name>`), awesome-claude-video (videos Claude made, with clips), opus-video-prompts (copy-ready prompts) | Template poster or preview clip |
yang0/handraw-style entries are style traits in Chinese (line, color, figure shape) with a sample, not scene prompts: keep the traits, add the user's subject, and describe the look instead of naming the reference artist in its tags.
Gaps: no structured library covers Midjourney, Flux or Seedream images, or Runway and Wan 2.x clips well. For those, search the nearest kind without `--model`, then keep the user's target model as the constraint: adapt with the model's row in `image.md` or `model-video.md`, and when the target has no row there, check the vendor's official docs for its parameters before adapting (mark what you could not check `UNVERIFIED`). Say in the fit line that the samples come from another model, and never switch the user's channel silently.
## Model names users say
Pass `--model` only when the user names a model or the channel is fixed. Keys:
| User says | `--model` |
| ----------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| 即梦, Seedance, Dreamina | `seedance` |
| 可灵, Kling | `kling` |
| 海螺, Hailuo, MiniMax, H3 | `hailuo` |
| Veo, Google 视频 | `veo` |
| Grok, Grok Imagine | `grok-imagine` |
| 万相, Wan | `wan` |
| GPT Image, gpt-image, 4o 生图 | `gpt-image` |
| Nano Banana, Gemini 生图 | `nano-banana` |
| Midjourney, MJ | none: no library has Midjourney samples; search without `--model`. For images adapt with the Midjourney row in `image.md`; for Midjourney video, `model-video.md` has no row yet, so check the official docs for its video parameters and mark what you could not check `UNVERIFIED` |
| Hyperframes, Remotion, Manim | same name |
With `--model`, matching samples come first; when fewer than 3 exist, the rest are filled from other models and flagged `modelMismatch`. The gallery says so on the card.
## Ranking
Keyword match (rare words weigh more, title over description over body) → a small readiness prior (detailed prompt, several samples, playable clip for video, no placeholders) → the user's own scores from `rate`. Results are spread over sources (at most 2 per source unless `--source` or `--model` is given), so the 3 picks are less likely to be near copies.
## Freshness and first use
Each source is checked at most every 10 minutes with ETags. A call waits about 6 s for downloads; slower sources keep downloading in the background and the result lists them under `pending`. When `downloading` is not empty and fewer than 5 results came back, say once that more libraries are still downloading and rerun the same command with `--wait-sec 30`; `refreshing` sources already served their cached data, so do not wait for them. Offline, results come from the cache or the bundled seed (`origin: seed`). The seed holds prompt text and links: its sample images and clips are remote, so offline they do not show. Say which of three things work: finding prompts (yes, from cache or seed), seeing samples (only for cached media), generating (only with a local engine such as Hyperframes, or once online). Offer the text-only path: pick by the prompt and its fit line.
## Licenses
- `seed: "allowed"` sources (CC0, CC BY, MIT, Apache) may ship in `assets/seed-prompts.json`. All other sources are downloaded to the user's cache at runtime and never bundled.
- Show source and license on every candidate (the gallery does). Keep `author` and `link`: CC BY needs credit when the result is shared, and the original post is where the user checks the sample is real.
- Entries whose repo declares no license are reference only: adapt the technique, do not republish the prompt text.
## Adding a library the user found
When nothing fits and `access.onlineSearch` is true:
1. `studio.mjs github "<topic> prompts"` lists repositories.
2. Open the repo's README or data file and check that it has prompts with sample media and a license.
3. `studio.mjs sources add <owner/repo> --kind image|video|code-video` detects the format (JSON data file, one YAML per entry, or a README with prompt blocks), test-syncs it and prints the entry count and two sample titles. Pass `--adapter`, `--file` or `--map` only when detection fails. A source that yields 0 entries is not kept. For a video or code-video library, check that a sample entry has `videos`; when the data file holds only a poster and a slug but the clip sits at a predictable address, pass `--video-template 'https://host/{slug}/clip.mp4'` (any `{field}` of the row), and open one built URL to confirm it plays. Without it the gallery shows stills only.
4. User sources live in the cache (`user-sources.json`), not in the skill, and are searched like the built-in ones.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| canvas-designanthropics/skillsCreate beautiful visual art in .png and .pdf documents using design philosophy. You should use this skill when the user asks to create a poster, piece of art, design, or other static piece. Create original visual designs, never copying existing artists' work to avoid copyright violations. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| docxanthropics/skillsUse this skill whenever the user wants to create, read, edit, or manipulate Word documents (.docx files) or Word templates (.dotx files). Triggers include: any mention of 'Word doc', 'word document', '.docx', '.dotx', or requests to produce professional documents with formatting like tables of contents, headings, page numbers, or letterheads. Also use when extracting or reorganizing content from .docx or .dotx files, inserting or replacing images in documents, performing find-and-replace in Word files, working with tracked changes or comments, or converting content into a polished Word document. If the user asks for a 'report', 'memo', 'letter', 'template', or similar deliverable as a Word or .docx file, use this skill. Do NOT use for PDFs, spreadsheets, Google Docs, or general coding tasks unrelated to document generation. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| pdfanthropics/skillsUse this skill whenever the user wants to do anything with PDF files. This includes reading or extracting text/tables from PDFs, combining or merging multiple PDFs into one, splitting PDFs apart, rotating pages, adding watermarks, creating new PDFs, filling PDF forms, encrypting/decrypting PDFs, extracting images, and OCR on scanned PDFs to make them searchable. If the user mentions a .pdf file or asks to produce one, use this skill. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| pptxanthropics/skillsUse this skill any time a .pptx or .potx file is involved in any way — as input, output, or both. This includes: creating slide decks, pitch decks, or presentations; reading, parsing, or extracting text from any .pptx or .potx file (even if the extracted content will be used elsewhere, like in an email or summary); editing, modifying, or updating existing presentations; combining or splitting slide files; working with templates (.potx), layouts, speaker notes, or comments. Trigger whenever the user mentions "deck," "slides," "presentation," or references a .pptx or .potx filename, regardless of what they plan to do with the content afterward. If a .pptx or .potx file needs to be opened, created, or touched, use this skill. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |