slide-builder_skill

Overview

This skill transforms essay-to-speech output into presentation-ready slide decks and video-ready assets. It converts talk tracks, semantic tags, and image assessments into a Talk Track v5 markdown package and exportable outputs (HTML, Remotion video, PPT-ready assets). Use it to move from spoken content to polished slides quickly.

How this skill works

The skill parses essay-to-speech output by extracting sections (Original, Talk Track, Images, Slide Ideas) and recognizing semantic tags like [HOOK], [KEY_POINT], [EVIDENCE], and [CTA]. It maps tags to slide types, applies slide count heuristics and timing estimates, and processes image ratings (USE/ADAPT/RECREATE/SKIP) including generating Nano Banana prompts when needed. Final output is a Talk Track v5 markdown file with frontmatter, slide index, individual slides (with audio blocks) and timing guidance for TTS or video exports.

When to use it

• You have essay-to-speech output and need a slide deck quickly
• You want to convert a talk track or transcript into presentation slides
• You need a video-ready presentation with synced narration
• You want structured markdown (Talk Track v5) for downstream rendering
• You need image prompts generated for visuals that must be recreated

Best practices

• Feed complete essay-to-speech output including Images and Slide Ideas sections
• Keep one idea per slide; use the tag-to-slide mapping to avoid dense slides
• Aim for 45–60 seconds per slide for pacing and simple timing math
• Use RECREATE prompts with 16:9, large labels, and brand colors for clarity
• Review the generated slide plan before exporting to video or HTML

Example use cases

• Turn a recorded keynote talk track into a speaker-ready slide deck and narration
• Convert a training module transcript into timed slides for e-learning video
• Generate presentation markdown and Nano Banana image prompts for designers
• Produce HTML slides with speaker notes for in-person rehearsals
• Export synchronized audio and timing JSON for Remotion video rendering