Adopt DFC Bridge with one real file

Do not begin by renaming the archive or rebuilding the component library. Run a small pilot on one working page or one new section and measure whether the handoff became clearer.

Start at the selection boundary: the modal resolves inside it, but the /workshop route needs lookup outside it. That Deferred dependency makes this example Partial. A section check is not always Partial, and it does not certify the legacy host or runtime integration. Open full size.

For the preparation steps in Figma, use the visual introduction. This pilot measures whether those steps help a recipient understand a real file.

What the pilot should answer

• Can another person find the page and its widths?
• Can they understand actions, states, and content rules?
• Did the number of follow-up questions fall?
• Is the preparation effort worth repeating?

Choose a useful scope

Pick a page with real content, at least one responsive decision, and a meaningful interaction. Avoid a clean demo file with no states or changing data.

For a new section inside an existing product, mark the selected root instead of inventing page metadata:

checkout-summary [section=checkout-summary]

Keep the host page outside the first pilot. See the selected-section contract when the team needs the integration details.

Assign three roles

Role	Responsibility
Designer	Prepares the selected area and does not explain it during the handoff test.
Developer or reviewer	Reads the file, records questions, and runs the available check.
Owner or manager	Sets the success threshold and decides whether to continue.

One person may have more than one role, but the handoff test needs someone other than the author.

Run the pilot in five steps

1. Record the starting point

Ask the recipient to read the file before any changes. Record time, questions, and assumptions.

2. Add only missing meaning

Follow the designer quick start. Do not rewrite the archive or duplicate Figma properties.

3. Check the declared area

Run Check page for the page or Check selected section for the explicitly selected section roots. Fix relevant findings and rerun. Record unresolved dependencies with an owner. Treat the report as one input, not as proof that every product question is solved.

4. Repeat the handoff

Give the file to the recipient again without a call. Ask the same questions and record the same measurements.

5. Decide what to repeat

Keep only changes that reduced guessing. Classify remaining items as a blocker, an owned open question, an intentional exception, or a process improvement.

A practical success threshold

Start with these targets:

• the recipient answers the core questions in five minutes;
• no blocking issue remains in the declared area;
• every open question has an owner and a next date;
• the before/after time and question count are recorded.

Do not turn the pilot into migration

• Do not rename the whole library before measuring the first handoff.
• Do not make every warning a blocker.
• Do not hide real states or data to make the example look clean.
• Do not claim that one ready section makes the host page ready.

Continue with the designer checklist, full review, or roadmap only when the pilot shows a benefit.