Push a run to Figma

Put a capture run on its own Figma page, one clean frame per screenshot, ready to annotate.

VideoCapture run to Figma · captions on · transcript

Steps

  1. Review the run side by side, and reject any bad shot before it reaches Figma:

    node scripts/make-gallery.mjs --dir <out>/run1
    The run gallery: a dark page titled with the run ID, showing two screenshot cards with their IDs and HTTP 200 badges.
    The gallery for one run. Reject a bad shot here, before it reaches Figma.

  2. Plan the Figma page:

    node scripts/figma-plan.mjs --dir <out>/run1 --inventory <inventory> --config p1-editor.config.json --release <new version>

    It writes figma-plan.json and figma-page.js. It uploads nothing.

  3. With an AI agent that has the Figma connector (use_figma and upload_assets), or by hand:

    1. Run figma-page.js with use_figma. It creates the page RUN-<date> · <run id> and one empty, named rectangle per screenshot, and returns their node IDs. If a page with that name exists, it stops.
    2. Call upload_assets with those node IDs and scaleMode: "FIT". Save the result’s uploads to urls.json. The URLs are single-use and expire in 10 minutes.

    3. Upload the images:

      node scripts/figma-upload.mjs --dir <out>/run1 --urls urls.json

      It refuses any URL that isn’t https on figma.com, before sending anything.

    A Figma run page with the run header and one clean frame per screenshot.
    One Figma page per run, named RUN-date · run ID, with one clean frame per screenshot.

  4. Record where each screenshot lives:

    node scripts/inventory.mjs record-figma --id <id> --evidence uploaded --file-url <file url> --page-name "<run page>" --node-id <clean frame node> --inventory <inventory>
VideoThe run in Figma · captions on · transcript

Next: Annotate with the kit.

This site uses Just the Docs, a documentation theme for Jekyll.