Slides plan (no model)

#Goals

• Expanded slides view = full-width cards, top of summary.
• Click slide = seek video timestamp (no modal).
• Descriptions scale with length setting.
• Always show all slides (even if text missing).
• No model call for slide descriptions.

#Data sources

• Primary: transcript timed text (already available with timestamps).
• Secondary: OCR text from slides (truncate, selectable).
• Tertiary: empty description (still render card).

#Description generation (no model)

• For each slide timestamp t:
• Pull transcript segments within a time window around t.
• Concatenate into plain text (no bullets).
• If no transcript: use OCR text (trim).
• If neither: empty string.
• Always render all slide cards; missing text → show slide only.

#Length scaling

• Map summary length to per-slide target chars.
• Use existing length presets (short/medium/long/xl/xxl + custom):
• short: ~120 chars/slide
• medium: ~200 chars/slide
• long: ~320 chars/slide
• xl: ~480 chars/slide
• xxl: ~700 chars/slide
• custom: derive from maxCharacters (e.g. maxChars / min(slideCount, 10), clamp).
• Clamp per-slide text: [80, 900] chars.
• Window size should expand with length (e.g. 20s → 90s).

#UI behavior

• Side panel slide mode is slide-first:
• vertical full-width cards by default
• thumbnail + timestamp + text
• transcript/OCR text appears before slide images finish extracting
• No giant summary block under active slide cards.
• Slide click: seek only (no modal).
• OCR toggle appears near summarize control only when OCR is significant

(enough slides + total OCR chars); otherwise hide it.

#CLI

• summarize <url> --slides streams a short intro paragraph and then a continuous narrative with slide images inserted inline where [slide:N] markers appear.
• The model is responsible for inserting every slide marker in order; text length is still governed by --length.
• If inline images are unsupported, the CLI prints text-only output and notes how to export slides to disk.
• Timestamp links use OSC-8 when supported (YouTube/Vimeo/Loom/Dropbox).
• Progress line reports slide extraction steps (includes slide counts when available).
• summarize <url> --slides --extract prints the full timed transcript and inserts slide images inline at matching timestamps.
• summarize slides <url> extracts slides without summarizing (use --render auto|kitty|iterm for inline thumbnails).
• Defaults to writing images under ./slides/<sourceId>/ (override via --slides-dir / --output).

#Implementation notes

• Build slideDescriptions map in panel:
• Use summary.timedText when available.
• Split transcript into segments with timestamps (already in payload).
• Store per-slide text on client (no daemon model calls).
• Ensure summary cache keys untouched; only client-only rendering.
• Slide extraction downloads the media once for detect+extract; set SLIDES_EXTRACT_STREAM=1 to allow stream fallback (lower accuracy).

#Steps

1. Add slide-description builder in sidepanel using transcript timed text + OCR fallback.
2. Add length-based per-slide char budget and window sizing.
3. Render expanded card list with timestamps + text.
4. Remove modal; click = seek only.
5. Add tests for slide description + fallback.