papperrollinggery/Paperrolling-DIRcreative-SKILL / skills/dircreative
dircreative
VerifiedUse after explicit $dircreative for film concepts, stories, scripts, director or creative groups, shot/asset planning, reference-video analysis, and image/video prompts. Real media execution requires the corresponding authorization. Validated ADCO handoffs are supported. DIRcreative source maintenance is outside this Skill.
references/prompt-model.md
View on GitHub# Model Prompt Craft
Use for bounded image/video prompts and revisions. Return the copyable prompt,
then essential bindings. Do not imply generation or media exists.
## Content and structure
Use `prompt-structure.md` for scene-led structure: establish the relevant intent,
reference roles and stable facts once, then write the evolving events with their
camera, environment, performance and sound. Simple beats can remain natural
paragraphs; multi-shot work can use a compact context plus timeline. Retain
source-authored craft, not an automatic twelve-part cinematic prefix. The user
may request a structured quality block; `video-quality-prefix.md` then describes
how authored fields are carried once per submission. Image work keeps still-image
rules. Do not add missing fields merely to fill a format.
Keep a complete tested prompt as the authored baseline. Change only requested
or demonstrably conflicting content; do not automatically translate, expand or
add a style capsule. When an upload-order correction keeps exactly the same input set, renumber it
mechanically. If files were added, removed or changed, inspect those inputs and
reconcile their roles and temporal state; a screenshot order is not reference
suitability or proof of generation.
Resolve the relevant visible facts before surface adaptation:
- identity, dominant action, environment, spatial relations and object states;
- shot size, angle, lens, support, movement endpoints, focus and temporal beats;
- motivated light, material, atmosphere, palette and grade;
- each input's role: identity, environment, composition, literal frame, style,
motion or planning-only;
- preserve/change boundaries, allowed incidental changes and targeted avoids;
- desired sound versus its actual generation or finishing route.
Templates supply structure. Compare people, props, setting, camera, actions and
sound with current facts. Remove old content instead of turning it into negative
lists. If residue remains, use `prompt-contamination-guard`. Keep internal IDs,
QA/account notes and local paths outside copyable text; aliases bind real inputs.
## Adaptation boundaries
1. Keep the supplied model/version/surface. Verify relevant capabilities officially;
do not guess reference limits, duration, audio, editing or extension support.
2. Prioritize identity, action, camera and chronology. To fit a prompt budget,
remove repeated adjectives/negatives, secondary atmosphere and optional metadata;
retain unique design constraints. Split overloaded action chains rather than
hiding missing capacity in prose.
3. Storyboards remain planning truth, not substitutes for clean first/end frames.
Hand-drawn shot-order/action/camera references require the exact card's
`storyboard_reference: conditional_with_explicit_role_binding`.
`conditional_inferred_from_omni_reference` permits a labeled, unverified trial
from documented image input. Prevent copied panel borders, shot labels, arrows
and production marks; retain intentional in-world text/SUPER. UI screenshots
and management tables stay planning-only.
4. Never preserve and change the same property in one edit. Rights audits,
capability receipts, hashes and generation authorization belong to execution
or formal handoff, not ordinary prompt revision.
5. Seedance 2.5 formal work uses Studio and the current `mr-li-seedance-25` method.
Add `seedance-25-emotion-prompt` only for emotion/micro-performance-led work.
Small revisions stay Fast without claiming the full method; cards own surface
capabilities.
6. Bind one unified recurring-character master per unit: headed by default, or
a user-requested/evidence-triggered headless-safe derivative, never both.
Add detail sheets only for visible shot needs. `character-master-sheet.md`
owns layout, provenance and review; this layer binds the selected asset.
State a specific missing capability/reference only when it changes usability.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |