Embeds Design ProposalExploring alternative visual treatments for embedded content from other documents, replacing the current green left border approach.

    Problem Statement

      Our documents support embedding content from other documents (embeds). Currently, embeds are visually denoted with a green vertical bar on the left side of the content, similar to a blockquote style. This approach has three key problems:

        Looks like a blockquote — The green left border is visually identical to how blockquotes are styled across the web, confusing the semantic meaning of embedded content.

        Not informative about the source — There is no indication of where the content comes from. Users cannot see the source document title, author, or any attribution without interacting.

        Disrupts reading flow — The continuous left border, especially for longer embeds, creates visual noise and breaks the natural reading rhythm.

      Constraints

          Users can navigate to the source document and expand/collapse embeds

          Visual treatment should be the same in both reading and editing modes

          Embeds can be any length — from a single paragraph to an entire document

          Three existing view modes: Content (inline), Card (summary), and Comments

    Current Design

      The current implementation uses a 3px green left border (border-l-primary border-l-3 in Tailwind) applied via the EmbedWrapper component.

      The green bar provides a visual cue that content is different, but it fails to communicate what that content is or where it comes from. It also creates a visual pattern that's indistinguishable from blockquotes.

    Proposed Alternatives

      Option A — Attribution Header

        Replace the green bar with a compact header bar above the embedded content, showing source document info.

        1

        How it works

          A small header displays the source document's icon, title, author, and a collapse chevron

          The header is clickable — navigates to the source document

          The embedded content below has no border or special styling — it reads naturally

          The container uses a very subtle background (#FAFAFA) with rounded corners to define boundaries

          The expand/collapse chevron is integrated into the header

        Pros

          Clear source attribution — solves the "where is this from?" problem

          Doesn't look like a blockquote — distinct visual pattern

          Minimal reading disruption — content itself has no special styling

          Familiar pattern (Notion synced blocks, GitHub embeds, Slack previews)

          Natural home for the expand/collapse control

          Works well for any embed length

        Cons

          Takes vertical space for the header row

          Nested embeds could create visual clutter with stacked headers

      Option B — Background Tint + Inline Attribution

        Use a very subtle background color with a small source label to differentiate embedded content.

        1

        How it works

          A very soft green tint (brand color at ~5% opacity) fills the embed area

          A small "From: Document Title" label appears at the top in the brand color

          A soft rounded border defines the embed boundaries

          Content reads almost like normal text with just a hint of differentiation

        Pros

          Lightest visual weight of all options

          Content reads very naturally — minimal disruption

          The tint provides a subtle but persistent visual cue across any length

          Source attribution is present but unobtrusive

        Cons

          May be too subtle for some users to notice at a glance

          Background tint can clash with dark mode if not carefully calibrated

          Less discoverable than a header approach

          No natural place for expand/collapse controls

      Option C — Margin Indicator + Hover Attribution

        Place a small icon in the left margin with source info revealed on hover. Zero visual disruption to the content.

        How it works

          A small link/clip icon sits in the left gutter (where collapse chevrons already exist)

          The embedded content has no border, no background — visually identical to regular text

          Hovering over the margin icon reveals a dark tooltip showing: source title, author, and "Open source" link

          Clicking the icon navigates to the source document

        Pros

          Zero reading disruption — content is indistinguishable from native text when scanning

          Minimalist approach that respects the reading experience above all else

          Uses existing gutter space (where collapse chevrons already live)

        Cons

          Low discoverability — users must know to look for or hover the icon

          Not obvious that content is embedded until the user interacts

          May cause confusion when users copy text without realizing it's from another document

          Tooltip-based attribution is invisible to screen readers

      Option D — Bookend Dividers with Source Tag

        Use thin horizontal lines above and below the embed, with source attribution inline in the top divider.

        How it works

          A thin horizontal line appears above the embed with the source document title embedded inline

          The content between the lines has no special styling

          A simple thin closing line marks the end of the embed

          Clicking the source title navigates to the source document

        Pros

          Very clean, editorial feel (similar to citations in academic papers)

          Clear start and end boundaries without a continuous border

          Source attribution is visible without interaction

          Works well at any embed length — the lines don't create visual weight like a continuous bar

        Cons

          Horizontal lines add visual noise if many embeds are stacked consecutively

          The editorial style may feel out of place in a more casual document

          Less clear boundary for very short embeds (1-2 lines)

    Recommendation

      Option A (Attribution Header) offers the strongest balance across all

      three problems:

        Clear source attribution (solves problem 2)

        Distinct visual pattern that's not a blockquote (solves problem 1)

        Content itself has no special styling, preserving reading flow (solves problem 3)

      A possible hybrid approach (A + B) could combine the header for attribution with a very subtle background tint for longer embeds, making boundaries clearer when the content spans many paragraphs.

    Next Steps

      Review mockups and gather team feedback on preferred direction

      Build a prototype of the top 1-2 options in the actual editor

      Test with real document content at various embed lengths

      Evaluate dark mode behavior

      Ship as an experiment behind a feature flag