linkprint/local-ai-movie-workspace
Skill Claude CodeCodex
Create MiniMax H3 prompts for a cinematic original-character reveal split into two independent 15-second clips with an explicit visual, pose, ambience, and music seam contract. Use for dense shot-driven introductions that establish a character in Part 1 and escalate without repeating coverage in Part 2; do not use…
linkprint/local-ai-movie-workspace
Skill Claude CodeCodex
Submit and retrieve real MiniMax H3 audiovisual video jobs through the reservation-scoped movie-ai Broker. Select the general or adult PinkCherry content profile and the approved workflow preset, then render T2VA, I2VA, FL2VA, L2VA, or Ref2VA only after the user approves every final value. Do not use for prompt-only…
linkprint/local-ai-movie-workspace
Skill Claude CodeCodex
Create MiniMax H3 prompts for safe television-style water-obstacle comedy driven by a short list of mandatory story beats while leaving choreography, broadcast coverage, crowd reactions, and timing details open to model improvisation. Use for original adult contestants, padded courses, recoveries, splashes, and a…
linkprint/local-ai-movie-workspace
Skill Claude CodeCodex
Submit and retrieve real local HunyuanImage-3.0-Instruct text-to-image jobs through the reservation-scoped movie-ai Broker. Use only when the current user request explicitly names Hunyuan or 混元, not merely because a prompt is complex and not for prompt-only writing.
linkprint/local-ai-movie-workspace
Skill Claude CodeCodex
List and render prompt-only images with the approved remote ComfyUI style models through the reservation-scoped movie-ai Broker. Use when the user asks for image-style model selection or names one of the supported safetensors models; do not use for reference-image Kontext or Flux2 Klein composition.
linkprint/local-ai-movie-workspace
Skill Claude CodeCodex
Submit and retrieve fast local Z-Image-Turbo text-to-image jobs through the reservation-scoped movie-ai Broker. It is the default for every local-language-model image request unless the user explicitly requests Hunyuan, and is used from an OpenAI-model session only when the user explicitly requests Z-Image-Turbo.