---
id: "9b23ba78-db00-487b-acd7-2cf23a065016"
name: "Presentation Text Extraction"
micro_summary: "`extract_text()` exposes presentation text, slide numbers, and UI chrome; exact per-slide grouping still needs a verified parser because active-slide text can repeat."
status: "in_progress"
see_also: []
scope: "Read editable text from an authenticated Google Slides presentation through browser-harness and establish the boundary between raw accessibility extraction and exact per-slide grouping."
---
## Verified raw extraction

```bash
browser-harness <<'PY'
tabs = [t for t in list_tabs() if "docs.google.com/presentation/" in t.get("url", "")]
if not tabs:
    raise RuntimeError("No authenticated Google Slides tab is open")
switch_tab(tabs[0])
wait(1)
extract_text()
PY
```

Observed output includes editable text from all slides, slide ordinals, and Google Slides UI chrome. The active slide's text can appear more than once because both presentation accessibility content and the active editing surface are exposed.

This is a reliable raw capture, not yet a verified per-slide parser. Text baked into images is absent and requires OCR. Speaker-note extraction has not been verified.