multimodal

Interleaving image, audio, and text modalities in prompts.

SKILL.md

GitHub

Multimodal

Interleaving image, audio, and text modalities in prompts.

Core Guidelines

  • Alignment: Anchor visual/audio inputs to textual markers in prompt templates.
  • Formatting: Format multimodal requests according to API client limits.
  • Token Budgets: Calculate image/audio tokens correctly.