[Feature] Support pptx as an input format (pandoc >= 3.8.3) #55

Links

• Epic: Tracking: gaps surfaced by an agent-driven document workflow (from #40) #44
• Origin: @pbarone's inbound workflow
• Pandoc release: 3.8.3, 2025-12-01
• Blocked by: the minimum pandoc version issue
• Related: [P2] Expose pptx in both directions (input requires pandoc >= 3.8.3) #49

Summary

Support reading .pptx so inbound decks can be converted to markdown for an agent to reason over. Pandoc gained a native PowerPoint reader in 3.8.3:

Add pptx (PowerPoint) as new input format (Anton Antich).
New module Text.Pandoc.Readers.Pptx, exporting readPptx``

No non-pandoc dependency is required.

Motivation

A user's documented workflow converts inbound .docx and .pptx from other people into markdown so his agent can read them. The docx leg works. The pptx leg is unsupported by this server despite being supported by pandoc.

This is the highest-value missing reader for the inbound half of a document workflow. Slide decks are a common way technical content arrives from other people.

Requirements

• pptx accepted as input_format
• Gated on the declared minimum pandoc version, or on runtime capability detection
• A clear error when the local pandoc is too old, naming the detected version and the required one
• Media handling considered alongside the docx media issue, [P1] Expose extract_media so docx to markdown keeps embedded images #48. Decks are image-heavy and the same silent-drop behaviour likely applies
• Tests skipped rather than failed when the local pandoc predates 3.8.3
• README states the pandoc version required

Notes

Pandoc 3.8.3 also added an xlsx reader. Worth a separate decision, not folded in here.