Drama Storyteller · عوامی Open Skill
Drama Storyteller
Drama Storyteller Digen پر ایک عوامی Open Skill ہے۔ کھولیں اور دیکھیں یہ ورک فلو کیا تخلیق کرتا ہے۔
ShortDrama
5 1 5
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/4091957/2026-10-03/uny35yud.webp
# Drama Storyteller Turn a drama idea or character into a story in the format the user wants — from a text-only script up to a fully voiced, scored video episode. The default protagonist is **Little Peach** — a young woman who claims she's a billionaire (see `references/little-peach.md` for her full profile). The skill works for any drama idea, not just Little Peach. ## When to use - User gives a drama idea, character, or one-liner and wants a story or video episode. - User wants to continue an existing drama (next episode), rewrite the script, restyle a character, or re-render a single clip. - User references "Little Peach" or the drama they started before. ## Reference files - `references/little-peach.md`: Little Peach's character profile and the drama's default tone. Read it before writing a script when the user is working on Little Peach. ## Output formats (mixed) Ask the user which format they want, or infer it from their request. The skill can produce any of these: - **Text-only**: just the written script/story (no images, no video). - **Illustrated story**: the script plus scene images (no video/voiceover/music). - **Full video episode**: the script, images, video clips, voiceover, and music, mixed into a finished episode. - **Single clip / single asset**: just one scene image, one video clip, one voiceover line, or one music track. ## Steps 1. **Check memory first.** Call `memory_query` for anything the user has told us about this drama (Little Peach's backstory, preferred tone, episode progress, style). If found, honor it; don't invent facts that contradict stored memory. If the user is starting fresh, proceed with the idea they gave. 2. **Clarify the episode and format.** Ask (briefly) what the user wants: a new episode, a rewrite, a restyle, or a single clip — and which output format (text-only / illustrated / full video / single asset). Confirm the drama genre/tone (default: dramatic comedy — Little Peach fakes wealth while the truth threatens to unravel). Confirm orientation (default `portrait` 9:16 for short-form drama). 3. **Write the script.** Produce a short drama script (a few scenes, with dialogue and shot descriptions) and save it with `create_document(content=<markdown script>, name="<episode title> script", entity_type="scripts")`. Keep it tight — enough for 3–6 short clips. If the user chose **text-only**, stop here and present the script. 4. **Generate character sheets.** For each main character, generate a character identity image with `generate_image` (`model='t2i'`, `orientation='portrait'`), `name="<Character> sheet"`, `entity_type="characters"`. For Little Peach use the profile in the reference file. Show these to the user for approval before proceeding. (Skip if the user only wants text.) 5. **Generate scene images.** For each scene/shot in the script, generate a scene still with `generate_image` (`model='t2i'`, `orientation='portrait'`), `name="<episode> scene <n>"`, `entity_type="storyboard_images"`. Keep the character's look consistent with the approved sheet (describe the same outfit/hair/face). If the user chose **illustrated story**, stop here and present the script with its scene images. 6. **Generate video clips.** For each scene still, generate a short clip with `generate_video` (`model='i2v'`, `input_urls=[<scene image asset_id>]`, `orientation='portrait'`, ~4–6s), `name="<episode> clip <n>"`, `entity_type="storyboard_videos"`. For character-consistency shots, you may use `model='ref2v'` with the character sheet as a reference image. 7. **Generate voiceover.** For each character's dialogue, generate voiceover with `generate_tts` (`model='tts.design'`, `voice_instruction=<describe the voice>`, text = the dialogue line), `name="<Character> VO <n>"`, `entity_type="voiceovers"`. Give Little Peach a bright, confident, slightly theatrical voice. 8. **Generate music.** Generate a background score with `generate_music` (`model='music'`, prompt describing the dramatic-comedy mood), `