director

Directs and produces complete AI-generated videos from scripts through final clips.

769|106|Updated Jul 31, 2026
One-click install
npx skills add https://github.com/s1dashu/director --skill director-s1dashu
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: director
Source: https://github.com/s1dashu/director
Command: npx skills add https://github.com/s1dashu/director --skill director-s1dashu

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) and assets (resource) components.

What problem does it solve? Creating a complete AI-generated video involves far more than writing a single prompt: you need a script, a visual style, consistent characters and voices, shot-by-shot prompts, and tracked generation tasks. This Skill orchestrates that entire pipeline so a video idea, personal story, or finished screenplay becomes downloadable clips ready for editing. ## Core Features & Use Cases - Three validated production modes: Animated Explainer for knowledge topics, Storytime Animation for first-person personal stories, and Cinematic Drama for script-driven AI films and short dramas. - Character and voice consistency: Generates confirmed character reference images, multi-view character cards, and locked voice anchors (from a built-in WAV voice library or newly created voices) so identities stay stable across clips. - Prompt and asset engineering: Produces per-15-second-segment video prompts following Seedance 2.0 official guidance, with scene, shot, camera, and continuity rules saved as individual Markdown files. - Execution via CLI tools: Generates, tracks, and downloads media through the LibTV CLI or Jimeng (Dreamina) CLI, with task IDs, statuses, and local file delivery. - Use Case: Ask it to turn a finalized screenplay into an AI drama series: it splits the plot into 15-second segments, builds character cards and per-role voices, writes shot-level prompts, generates a representative clip for approval, then batch-generates and downloads all segments. ## Quick Start Use the director skill to create a two-minute animated explainer video about why people procrastinate, confirming the script and visual style with me first.

Frequently Asked Questions about director

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create a complete AI-generated video with an agent?

Start from a topic, personal story, or finished script and let the director skill route it to a validated mode. It handles scriptwriting, visual style selection, character and voice design, per-segment prompts, generation, and download of clips ready for editing.

How do I keep characters consistent across AI video clips?

Generate a confirmed character reference image for each recurring character before video generation, then connect only the references each segment needs. Cinematic Drama uses 16:9 multi-view character cards derived from confirmed candidates via image-to-image generation.

What video generation models and tools does it support?

It targets Seedance 2.0 (Pro at 720p for Cinematic Drama) executed through the LibTV CLI, with the Jimeng (Dreamina) CLI as an adapted alternative. Model names, parameters, and input limits are always verified against the live CLI schema before generation.

How do I keep the same voice across all video segments?

Choose a voice from the built-in WAV voice library or create a new one from segment 1, then lock it as a voice anchor. All later segments connect the same standardized audio as a timbre reference while each prompt supplies its own exact narration text.

Can it write a script or build a world from scratch for Cinematic Drama?

No. Cinematic Drama requires an already-confirmed worldview, character settings, and screenplay; it only performs production breakdown into segments and prompts. For concept-driven videos without a script, use Animated Explainer or Storytime Animation instead.

Why does my generated video have audio glitches at the end?

Generative video models occasionally produce short audio artifacts in the final fraction of a second. Keep narration length within the recommended limits and trim roughly the last 0.1 seconds of each clip in an editing tool after downloading all segments.