lora-scene-partner

LoRA scene partner establishes proactive collaboration for multi-agent LoRA prompt orchestration in video-generation workflows like Graffito and Akira. It manages persona, HARD/SOFT context constraints, and command semantics (TACTIC, DIRECTION, FLUSH, RESET, META) to coordinate prompts across the prompt-engineering-toolkit, creative-lens-toolkit, and img2vid-pipeline while preserving consistency.

HARD and SOFT context management tracks rigid constraints versus flexible parameters in LoRA-based scene generation. HARD constraints enforce non-negotiable requirements; SOFT constraints allow variation. LoRA scene partner enforces these distinctions alongside FLUSH/RESET semantics to maintain control and adaptability across collaboration cycles.

Yes. The quick-start workflow applies LoRA scene partner to baseline prompts by configuring partner persona and running a collaboration cycle with prompt-engineering-toolkit to iteratively refine scenes before passing to video-generation pipelines, ensuring constraint-informed creative output.

STANCE CHECKs verify alignment across prompt-engineering, creative-lens tooling, and Img2Vid pipelines. Self-Check Cadence enforces periodic validation of persona and context state. Together they preserve SPINE (core intent) over MIRROR (surface variation), keeping multi-agent orchestration coherent throughout generation cycles.

FLUSH clears temporary or session-scoped context while preserving persona and HARD constraints. RESET reinitializes the entire collaboration state. LoRA scene partner enforces both semantics to let you control how much collaboration history carries forward into the next generation cycle.

Yes. LoRA scene partner routes prompts to prompt-engineering-toolkit, creative-lens-toolkit, and img2vid-pipeline while enforcing command semantics, persona constraints, and context rules. It serves as an orchestration shell that preserves intent consistency across downstream toolkit execution.