Omni-Rewriter can validate LingBot-Video captions and call scripts/inference.py

[WayneJin0918]

Hi — Omni-Rewriter is an open prompt-expansion harness. We added a typed LingBotCaption schema plus a bounded subprocess runner around LingBot-Video’s public inference script (scripts/inference.py), so expand and generate stay separate.

Scope

• We do not claim LingBot-Video implements /v1/videos.
• The runner only exposes a fixed CLI surface (mode, backend, size, steps, seed, optional image / refiner).
• Paths come from LINGBOT_VIDEO_ROOT / MODEL_DIR — no machine-specific defaults in the repo.

Links

• Repo: https://github.com/WayneJin0918/Omni-Rewriter
• Adapter notes: https://github.com/WayneJin0918/Omni-Rewriter/blob/main/docs/dialects/generation-adapters.md
• Replay / PE gallery: https://huggingface.co/spaces/Wayne-King/omni-rewriter-replay

FYI only — not a request to change lingbot-video. Happy to correct any contract misread.