AI Drama Series Studio — Open Skill
AI Drama Series Studio
Drama Series Studio produces short vertical drama series end-to-end. Two build modes, chosen once per series: - Auto: a few lines + a character reference image -> the skill writes the story bible, script, and every image/video prompt itself. - Manual: you supply the story bible, character refs, props, and every image/video prompt -- the skill executes and only asks when something is missing. Continuity system: - Visual DNA: every character reference is transcribed into literal face/skin/hair/outfit/background details (not invented atmosphere) and passed into every relevant scene generation call. - Automated drift detection: each generated scene keyframe is checked against the Visual DNA and likely mismatches are flagged before you review. - Group-and-split option: when separately generated characters do not match in style or lighting, the cast can be generated together in one image and split into individual refs. Two-stage approval: 1. Character references are approved first, as their own gate. 2. A full episode's scene images are generated and reviewed together as a batch, not one at a time -- side-by-side comparison is what actually catches drift. Other details: - Aspect ratio (9:16 or 16:9) is a per-series choice, applied consistently across every generation call. - Hybrid voiceover: a narrator carries most of the story, with characters speaking select lines at key beats; narrator gender is a per-series choice. - Cast defaults to African-American, using descriptive (not racial-label) skin-tone language rotated across the cast. - The available image model is checked once per new series rather than assuming a fixed model. - The finished episode is exported as a real video file, and platform-fitted social captions are generated as the final step. - A character's stored data can be cleared and reset before regenerating a redesigned version of them.
ShortDrama
64 6 20
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/925535/2026-09-25/8w4z4hqn.png
--- name: drama-series-studio display_name: AI Drama Series Studio description: Create realistic AI drama scenes, dynamic dialogues, and consistent characters — auto-write a full series from a one-line idea, or execute exactly from your own story bible, prompts, and reference images. Default cast is African-American; you choose vertical or landscape. allowed-tools: - generate_image - generate_video - generate_tts - generate_music - mix_clips - create_document - get_document - update_document - memory_query - memory_store - list_models - recognize_image - split_image - download_video - memory_forget --- # Drama Series Studio ## When to use - User wants a **long AI drama series** (multi-episode, ~2-minute episodes, built from 8–12 second scenes) for social/YouTube — either built entirely from a one-line idea, or built exactly to the user's own script/prompts/reference images. - User wants to **continue** the series with the next episode, revise a script, or redo one scene/clip. ## Fixed & user-set preferences - **Default cast:** African-American characters unless the user specifies a different cast for a given series. This means contemporary Black American — not continental African, not a diaspora/heritage story — unless the user explicitly asks for that: - **Names:** contemporary American names, drawn from both classic American names and names distinctly common in Black American communities (e.g. Michael, Angela, Denise, David, Marcus, Jasmine, Aaliyah, DeShawn, Destiny, Andre, Imani). Do **not** default to continental African given names (Kofi, Amara, Kwame, etc.) or invented-sounding names — those read as African, not African-American, and are a different identity. - **Setting:** a contemporary US city, suburb, or town by default, not an unspecified coastal/fantasy port that reads as generically African or Caribbean — unless the story explicitly calls for a different setting. - **Style/hair:** contemporary Black American fashion and hairstyles — box braids, locs, silk press, waves, fades, streetwear or everyday contemporary American clothing — rather than continental African garments (kente cloth, dashiki, traditional headwraps) unless the user is specifically telling a heritage/diaspora story. - **Skin tone:** describe with layered descriptive terms, never racial labels — rotate across the cast rather than defaulting everyone to one tone: deep ebony skin, rich dark brown skin, deep mahogany skin, warm brown skin, warm cocoa complexion, caramel complexion, honey brown skin, golden brown skin. - **Build mode — ask once per series:** **Auto** or **Manual** (see below). Don't assume; a returning user may want either mode episode to episode. - **Orientation — ask once per series:** **9:16 vertical** (TikTok / Instagram Reels / YouTube Shorts) or **16:9 landscape** (YouTube long-form). Apply the same orientation to every `generate_image`, `generate_video`, and `mix_clips` call for that series. - **Episode length:** ~2 minutes (≈10–12 scenes of 8–12 seconds) by default — this is the cadence that has worked well before. Confirm if the user wants something shorter/longer. - **Episode count:** flexible — ask; if unspecified, propose a plan (e.g. 5–10 episodes) and confirm before building. - **Voiceover — hybrid narrator + dialogue:** the narrator carries most of the story; characters speak select lines directly, used for key emotional beats or punchlines rather than every scene. Ask the user once per series whether the **narrator voice is male or female**; keep it consistent for the whole series unless they change it. - **Genre/tone:** don't hardcode a genre — derive it from the user's idea (Auto mode) or from their story bible (Manual mode), and confirm it back to them. - **Language:** communicate with the user and write all story/script/caption content in English. - Keep content family-friendly by default; if the user asks for adult/NSFW content, politely decline and say it's out of scope for this skill. ## Build modes