Pikapika260214/rw-consulting-ppt

What it solves

This project addresses the difficulty of creating high-quality consulting-style presentations when the input material is raw or unstructured. Instead of focusing on simple aesthetic beautification, it transforms rough industry reports, meeting minutes, and interview notes into structured, "proof-object-first" consulting slides that follow a professional storyline and evidence structure.

How it works

The project consists of two independent Codex skills:

1. rw-consulting-ppt: This skill follows a rigorous dialogue-based workflow (Alignment $\rightarrow$ Input Synthesis $\rightarrow$ Blueprint $\rightarrow$ Sample Brief $\rightarrow$ Sample Gate $\rightarrow$ Batch Generation). It generates high-density 16:9 PNG images of slides using "proof objects" (like funnels, route maps, and capability stacks) to carry the argument, then packages them into an image-only PPTX file.
2. ppt-to-editable: This skill converts image-based slides (PNG/JPG/screenshots) or image-only PPTX files back into editable PowerPoint presentations. It uses a Python-based pipeline involving OCR, background cleaning, and reconstruction of native PowerPoint shapes and tables to restore editability where possible.

Who it’s for

• Strategic Analysts: Those who need to synthesize complex industry research into management-level decks.
• Consultants: Professionals creating recap decks from meeting transcripts or brainstorm sessions.
• Business Leaders: Users who want to transform rough business judgments into professional, evidence-based visual presentations.

Highlights

• Proof-Object-First Approach: Prioritizes visual structures that prove a point over simple bullet points or templates.
• Rigorous Quality Gate: Requires sample page approval before batch generation to prevent wasted tokens and incorrect styles.
• Two-Mode Conversion: Offers a "token-saving single-page mode" and a "high-quality multi-agent mode" for restoring editability to image slides.
• Storyline-Driven: Forces the definition of a working thesis and page-level claims before any design begins.