Canva Presentation Time Fitting

Presentation Time-Fitting

Take an existing Canva presentation and align it with a target speaking duration using presenter notes only. Visible on-slide text is read for context and timing, but must not be changed by this skill. The flow is: confirm design + total duration (before cloning), read slide text, divide time across slides, add, replace, or trim speaker notes per slide to match each slot, write notes only back via Canva editing tools.

Primary workflow (time-balanced presenter notes)

Step 1: Collect design and duration (before other Canva tool calls)

• Gather everything you need before calling Canva:resize-design, Canva:get-design-content or other editing/content tools:
  • Design: link (canva.link), full Canva URL, design ID, or design name (see Step 2 for how each maps to design_id)
  • Total speaking time for delivering the deck (e.g. "15 minutes", "45 min", "1 hour"). Parse to total seconds (or minutes as a decimal). If the user is vague, confirm the total minutes they want for speaking through the slides (include Q&A in that total only if they say so)
• If either design or duration is missing, ask and wait before proceeding

Step 2: Resolve the design

• If the user provides a short link (canva.link), call Canva:resolve-shortlink to get the design URL
• If the user provides a full Canva URL, extract the design ID (https://www.canva.com/design/{design_id}/...)
• If the user provides a design ID directly (typically starts with D, e.g. DABcd1234ef), use it as design_id; do not use Canva:search-designs for a raw ID
• If the user provides a design name (not an ID), use Canva:search-designs; if multiple matches, ask for clarification
• If you still cannot resolve a design, ask for the design ID or link

Step 3: Work on a copy (required)

• Use Canva:resize-design with the same dimensions as the source to clone the presentation; do not modify the original

Step 4: Read slide content

• Call Canva:get-design-content on the copy to extract visible text per slide/page (headings, bullets, body)
• Build a per-slide summary: slide index or title, text content, and whether speaker notes already exist if the payload includes them
• Slide count = number of content-bearing pages from this response (use for the division in Step 5)

Step 5: Calculate time per slide

• Target speaking time per slide = total_duration ÷ number_of_slides (equal split unless the user asks to weight title/closing differently; then adjust and state the assumption)
• Derive a word budget per slide using ~150 words/minute as a guide: target_words ≈ (target_minutes_per_slide) × 150 (floor/round sensibly; treat as guidance, not a hard cap)
• Show the user a small table: slide count, total duration, seconds (or mm:ss) per slide, approximate word budget per slide

Step 6: Generate speaker notes (in this session)

• Generate the notes directly in this response; no external API call is required. For each slide, using that slide’s text from Step 4 plus the per-slide time/word budget from Step 5, compose presenter notes (this is the only content you may author or edit for timing):
  • Aim to fill roughly the allocated time when spoken aloud
  • Expand thin slides with context, transitions, and delivery cues in the notes; shorten notes if the budget is small (do not shorten on-slide text)
  • Stay faithful to the slide’s visible messages; do not invent facts
• If a slide is visual-heavy with little text, write notes that describe what to say about the visual in the allotted time

Step 7: Write notes back to Canva

• Call Canva:get-design-pages on the copy to obtain page_id (or equivalent) for each slide that needs notes
• Call Canva:start-editing-transaction on the copy if not already active; use the returned transaction_id for edits
• Optional ordering: start the editing transaction soon after you have the copy’s design ID (e.g. right after Step 4 or right after get-design-pages) so the session is open before you spend a long time composing notes in chat; some environments time out less often when the transaction is not deferred. If the connector allows one long-lived transaction for the whole flow, prefer opening it once and reusing transaction_id for Step 7 edits
• Call Canva:perform-editing-operations using only the operation type the Canva MCP exposes for presenter/speaker notes (name may vary by connector; use the schema for updating notes per page). Do not call replace_text, find_and_replace_text, or any operation that changes visible on-slide text. Apply all note updates, batched in as few calls as allowed
• Call Canva:commit-editing-transaction to save