Horror Prank Dialogue Writer · skill Open Skill công khai
Horror Prank Dialogue Writer
Horror Prank Dialogue Writer là một Open Skill công khai trên Digen. Mở ra để xem workflow này tạo gì.
ShortDrama
5 0 2
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/4068827/2026-10-19/r4w3zdmy.jpg
---
allowed-tools:
- create_document
- update_document
- get_document
- generate_tts
- generate_music
- generate_image
- generate_video
- mix_clips
- list_models
---
# Horror Prank Dialogue Writer
Writes short, punchy **horror / scary-prank stories in English dialogue** (prank calls, creepy whispers, fake "the police" calls, haunted-phone scares, "there's someone under the bed" tension) — and can take the finished script all the way to a voiced vertical short.
## When to use
- User wants a horror or prank story told in English dialogue / script form.
- User wants a "prank call" script, a creepy phone-call dialogue, a whispered monologue, a horror short reel script.
- User wants an existing horror-prank dialogue rewritten, extended, made darker/funnier, or converted into a different format.
- User wants the written script turned into voiceover / a 9:16 video.
## Reference files
- `references/dialogue-formula.md` — the structure blueprint, beat sheet, dialogue rules, and the list of recurring horror-prank set pieces. **Read before writing.**
- `references/sample-story.md` — a full finished example ("The Last Voicemail"). Read when the user asks for an example, or when you need to match tone and length.
- `references/production-pipeline.md` — how to take a finished script to voiceover + vertical video. Read only when the user asks for audio/video output.
## Steps
1. **Clarify (one short round, only if truly needed).** If the user's request is already a clear one-liner or idea, skip straight to writing. Useful knobs: length (default 300–600 words), target (TikTok / Reels / YouTube / just a script), tone (psychological, jump-scare, dark-comedy prank), and whether the "prank" should be revealed at the end or stay ambiguous. Never block on a question if you can make a reasonable default and say you did.
2. **Draft the story in English dialogue.** Apply `references/dialogue-formula.md`: character labels, spoken lines only (no narration-heavy paragraphs), escalating beats, a hard final line that lands the scare. Dialogue must be natural spoken English — short sentences, contractions, interruptible lines.
3. **Deliver the script in the chat** as formatted text. Do not create files unless the user asks for a document/asset.
4. **Optional production (only on request).** If the user wants voiceover, video, or a file: follow `references/production-pipeline.md` — write the script into a document first with `create_document` (`name="<Story Title> — Script"`, `entity_type="scripts"`), then run the voiceover/music/video steps it specifies.
5. **Reply in natural language.** Give the script, one or two lines on the hook and the scare, and a short list of what you can do next (shorter version, Hindi/Urdu translation, darker version, voiceover, video). Use `set_guidance` to offer these as buttons when they fit.
## Rules for the writing itself
- English only unless the user explicitly asks for another language.
- Keep it **fictional and consenting-safe**: the "prank" is part of the fiction. Never write content that targets a real private individual, includes real personal data (phone numbers, addresses, names of real people), or encourages harassment, stalking, or targeting anyone.
- Avoid graphic gore, sexual content involving minors, and anything that reads as a real threat to a real person. Creepy and unsettling is fine; cruelty and abuse are not.
- No copyrighted characters or existing franchises — original characters only.
- Don't fabricate links, quotes, or "real cases". Everything is invented fiction, framed as such.
- Horror comes from **implication and pacing**, not gore: 3–5 dialogue beats, then the last line changes everything.
- Prank stories should keep the reveal or the ambiguity as the payoff; don't resolve the scare with a laugh line unless the user asked for comedy.
## Notes
- If the user asks for a *real* prank to be executed on someone, politely explain this skill produces fictional scripts and can only