# Campaign page layout

Use this pattern for a marketing system with campaigns and deliverables. A solo
calendar request can stay smaller. Use a narrow navigation column with a next
action, tinted section headings, and visual campaign cards.

## First screen and connected destinations

Use a full-width home, roughly one-quarter navigation and three-quarters working
area. Put Campaigns, Content, and a substantive Brand kit in the navigation column, followed by
one contextual next action linking directly to its draft. Keep capture guidance
collapsed or use real configured creation controls; never draw fake buttons.

The main area starts with active campaign gallery cards. Use a coherent palette
and meaningful colored icons: campaign/launch, article, video, email, brand.
Show title, lifecycle state, planned launch, and lead; avoid redundant labels.
Use small cards at ordinary desktop widths so a handful of campaigns can share a
row. Wrap titles and useful summaries. Follow with a compact filtered editorial
review list. Keep full production boards and publishing calendars in Content.

When the user requests or welcomes a cover and original creative, deliver that
visual layer: a restrained brand-relevant cover and coordinated campaign title
treatments derived from the actual messages. Existing assets are preferable; their
absence is not a reason to omit authorized original design. Create original
artwork with an available image/vector tool, attach it, and use it as the campaign
gallery preview. Keep text readable at small card size, and distinguish proposed
creative from approved assets. The cover must not displace primary work beyond
the first screen. If artwork creation or attachment is unavailable, record that
specific gap instead of counting a text-only gallery as visual parity. For briefs
that do not call for creative, choose imagery only when it helps recognition. Never
fabricate product screenshots or use random photos to fill deliverable cards.

## Campaign briefs and content drafts

Campaigns are database records, not loose documents. Pin stage, launch and lead
near the title. Put the promise, audience/action, and current decision first;
then show Content filtered by the exact campaign relation. Keep large creative
in a collapsed section or separate asset area, not ahead of the working brief.
Show undated ideas in this view too. Shared Content offers production, publishing,
and unscheduled views, with each record returning to its campaign.

Write useful format-specific bodies from the supplied brief: an article draft
with editorial checks, a video with timed beats, an email with subject and body.
Use different reading structures for each. Brand guidance belongs in its own
readable destination and includes voice, message, and reusable creative direction.

## Check navigation and usability

At normal zoom with sidebar visible, identify the next campaign, its state/date,
and the next editorial action without scrolling through an introduction. Open a
card, find the brief and its deliverables, then open an article and a second
format. Verify the related-content filter excludes a second campaign and includes
undated ideas. Review that the cover and cards support this path rather than
consuming the screen. The artwork can vary; these capabilities must survive.

## Render and inspect the saved result

Build records and views first, then compose their existing blocks into the home.
Read the connection's current Markdown and view/layout documentation. For native
toggles and tabs, separate their label from child blocks with a blank line.
Preserve exact child-page/database tags during edits. A tool's success response
does not establish that moved blocks retained the intended order: inspect the
saved page and correct misplaced sections while preserving existing objects.
If moving existing database blocks into a new tab/column container changes sibling
order, fetch the saved hierarchy, then reapply the intended complete composition
using those same object IDs. Verify the second readback before deciding the native
layout is unavailable; do not discard working tabs merely because their first
move reordered the page.

Set width, card size, wrapped content and property visibility deliberately.
Check data-source update capabilities for database-record layouts, not just page
update capabilities. On connections exposing `page_layout`, read
`notion://docs/database-page-layouts`: pinned properties, sidebar details and
record full width can be configured there before UI inspection. Use
available UI controls for settings absent from MCP; do not simply describe how
the user could finish a supported layout. Hide duplicate linked-database titles.
A tab intended as a working table must contain an inline database/view, not merely
a link to its full-page source. In Notion Markdown, verify `inline="true"` for
embedded work surfaces; reserve `inline="false"` for navigation destinations.
Use semantic colored page icons so mentions are recognizable, not a repeated
colored dot. Keep original assets and long detail accessible without dominating
the working first screen.

Verify home → record → related work → home, actual view filters and counts, and
at least one alternate view. Restore the intended everyday default view after
testing alternates so delivery opens on useful work. Inspect the rendered result at normal zoom with the
sidebar visible. Compare hierarchy, density and navigation against the selected
reference. Check useful work is visible above the fold, long titles wrap, and
essential columns do not overflow. Correct failures before delivery. If rendering
or a capability is unavailable, report that exact gap without claiming parity.

During skill evaluation, preserve the first result and the run's evidence. If
review finds a miss, change the reusable instructions and rerun or resume the
skill with only that revised guidance. Reviewer hand-polishing is not a passing
skill test. Equivalent layout and workflow matter more than identical wording,
page IDs, image pixels, or arbitrary record counts.
