DR.PROMPT / CREATIVE ENGINE
Turn visual intent into production-ready prompts for any style and scene.
01 / CREATIVE ENGINE
Turn one rough idea into a production-ready video prompt through the full Creative Engine workflow.
02 / CREATIVE ENGINE
Professional workflows, cutting-edge expertise, and extensive real-world testing—delivering video prompts engineered for exceptional results.
03 / PROMPT LIBRARY
Ready-to-use cinematic prompt templates. Library covers action, cinematography, VFX, advertising, narrative and visual style. Continuously updated...
Select any node to inspect review details
Master Prompts
Choose a template for the current idea, then adjust characters, settings, actions, and shot details on the right.
Coming Soon
Unlock your creative potential.
For the first phase, structured storage such as JSON, Airtable, Notion Database, or Supabase keeps fields clear and maintainable.
{
"id": "ultimate-move",
"tag": "Signature move",
"triggerKeywords": ["signature move", "summon", "finisher"],
"scaffold": "charge -> summon -> finisher -> aftermath",
"shotGrammar": ["low-angle rise", "rapid push-in", "hold on the landing"],
"durationAdvice": "8-15 sec",
"executionNotes": ["make the action chain explicit", "specify the group count"]
}
When the user describes a signature move that summons five dragons to defeat a villain, the system matches Signature move、group summon、one-hit finisher and continues to drafting and review.
This is not a database implementation; it is a practical first-phase template asset schema that can later map to JSON, Supabase, or Airtable.
RAG should be a full loop of collection, structuring, review, publishing, and feedback—not just a static word list.
If this becomes a full product later, separate template assets, case feedback, and review standards into three admin views.
The review layer evaluates whether a draft deserves final refinement using explicit dimensions, weights, and pass signals.
Workflow Skills
Each module lists its role, boundaries, inputs, outputs, rules, and exclusions for future model swaps and prompt configuration.
Short-form creators, emerging commercial-video teams, cohort learners, and AIGC teams validating ideas quickly.
Ideas are hard to deconstruct, shot language is hard to write, outputs are inconsistent, and duration-to-shot planning is difficult to control.
Translate ideas and reference images into executable cinematic prompts.
The current build focuses on one prompt-generation path and avoids unnecessary multi-model branching.
Show the final prompt together with key requirements, matched templates, review results, and rewrite notes.
RAG should remain a maintainable structured template asset, not a pile of example prompts.
Text + references + duration + shot preference
Capture key requirements and creative focus
Decide single take / multi-shot / rhythm
Match a structured template from RAG
Draft, review, and rewrite
Final prompt + review summary + explainable workflow
The same API key powers text and reference understanding, director planning, prompt writing, review, and final rewriting.
API Contract
The current build exposes seven main workflow APIs and one multi-skill review call. Each module can use a separate model later without rewriting the workflow.
Checking the local proxy and text-model configuration.
V1 focuses on the text-generation workflow. The seven core modules are exposed as independent APIs, while review runs as one multi-skill call.
This section is for development diagnostics, not a core end-user feature.
{}
LOCAL SERVICE REQUIRED
This page was opened as a local HTML file and is not connected to the Dr.Prompt local service, so model settings cannot be saved or tested.
Windows: Extract the release package and double-click the Dr.Prompt launcher.
macOS: Open the DMG, drag the Dr.Prompt app to Applications, then launch it.
LOCAL SERVICE CLOSED
The local background service has stopped. You can safely close this page.