# Design the project home

Read this before composing or polishing the output. Treat the layouts as starting
points; adapt them to the user's content and working rhythm.

## Write for the project team

Use the project’s actual name and familiar labels: “Tasks,” “Docs & decisions,”
“Meetings,” and “Overview.” Prefer “Target date” to an unexplained “Target,” and
“Next milestone” to “Checkpoint” when introducing a new label. Keep the team’s
established terms and status values. Do not use “EPD,” “canonical,” or “relation”
as setup instructions for the reader.

Ask “How often does the team meet?” rather than “What is your cadence?” Call
reusable agenda pages “Reusable agendas,” and explain how to use them without
calling ordinary pages native database templates. In project summaries, describe
what success looks like instead of referring to an “acceptance approach.” Keep
technical acceptance criteria where they help the team do the work.

## Compose an EPD project home

- Start with the actual project name and one sentence about the user problem.
  Show stage, known leads, and agreed target in a compact metadata area. Use
  ordinary labels such as Overview, Tasks, Docs, and Decisions.
- Keep the overview brief: intended outcome, scope boundaries, and the current
  risk or decision. Link the product brief for details. Do not repeat the same
  summary in several panels.
- Give Tasks and Docs equal structural importance. Place a compact, populated
  task table and docs list/table on the home. Keep alternate status and review
  views in tabs on those collections. Do not turn every collection into a board.
- Put detailed requirements, design states, tradeoffs, decisions, and rollout
  criteria inside the relevant documents. Tasks link to their specifications;
  readers can return to the project home from either kind of record.
- Use restrained typography, whitespace, alignment, and consistent view settings
  for polish. Color should communicate state or draw attention to a real risk.
  Avoid promotional taglines, decorative phase cards, motivational copy, rainbow
  panels, and repeated emoji headings. A cover is optional, not the design.
- Include milestones only when they represent real project commitments. Do not
  invent a phase sequence to fill the page. Put detailed sequencing in tasks or
  the rollout doc instead of adding another home-page module.

The final project home is the body of a record in the chosen Projects database; the
initial version can be a private draft while placement is unresolved. Surface
existing project properties instead of duplicating editable status and ownership
in body text. Once placement is known, filter Tasks and Docs views to that project
relation. Keep team context compact: names, responsibilities, and relevant
review roles. Add it from known context or the user's post-delivery answers,
without an empty roster or fabricated assignments.

Compose visible metadata → milestone and team → tabbed Tasks, Docs, and Meeting
notes. Fit the first actionable task into the initial screen with the context
needed to act on it. Keep expanded scope in Overview. Show a compact notes view
and clear next-agenda entry; keep the full agenda inside its record. Adapt to the
team's context; a personal checklist or event brief can remain smaller. Use
connected records, clear states, and carefully chosen density for visual character.

## Keep the project home focused

Set the page to full width for dense team work. Keep the overview short
enough that work is visible without scrolling through a long introduction. Use
compact properties and view tabs rather than many colored summary panels.

## Build the connected pages

Create the real destinations and records, then compose the home around their
actual URLs and views. Embed the main working view in the home; a full-page
database link alone is not equivalent. Move the
view into its intended position if the tool initially appends it. Keep source
briefs and evaluation notes behind a secondary link or collapsed section so they
do not dominate the product experience.

No “notes will live here,” future-tense descriptions of missing features, fake
buttons, empty titled subpages, or views that have only been named but not
configured. With genuinely sparse input, deliver a smaller completed artifact:
useful capture prompts and honest empty states are acceptable. Do not manufacture
facts. When the user requests a demo, use explicitly labeled mock inputs as the
source; do not misrepresent them as real work or assign real people.

Explain how to use an unfamiliar view or section in one practical sentence next
to it. Distinguish editable prompts from committed tasks and static summaries
from live data. Avoid duplicating the same tasks as independent checklists and
database records without explaining which is authoritative.

## Review the saved result

Review the artifact itself, not its Markdown source or the final response:

- **Distinctive composition:** does the arrangement fit this use case, and does
  the rendered result make the next action clear? Changing the
  title and emoji on the same generic page does not pass.
- **Visual coherence:** do color, typography, icons, imagery, and spacing feel
  intentional across the home and child pages? Plain can be polished; random
  decoration and bare boilerplate both fail.
- **Immediate utility:** can a new user start doing the work from this page?
  Check a real item, its detail, and the return path. Verify action
  configuration without triggering unrelated writes.
- **Depth and fidelity:** are records populated with the supplied context and
  useful structure, while preserving uncertainty, blockers, and provenance?
- **Usability:** inspect the main surface on desktop and at a narrow width when
  rendering is available. Check for clipping, stranded headings, confusing
  reading order, excessive empty space, and missing undated work.

Revise a failed result before presenting it as finished. Record structural
validation and visual inspection separately.
Do not claim visual quality was verified by frontmatter, lint, or schema checks.
