Ready-made automation. A workflow runs inside the app on every picture a source delivers; a script drives the app from outside, and can do anything the app can.
Every method a script can call is in the API documentation
Opens a folder, waits until the app has measured every frame where the camera focused, then gives the sharpest tenth five stars, the next fifth three stars, and rejects the softest third. A start to look through, not a verdict.
bun auto-cull.ts <folder>Put the card in and cull: every picture you pick is sent to the newspaper's server as a JPEG developed from the RAW, with your by-line, credit, the event, the place and a caption template in it.
bun card-to-desk.ts <desk connection> [<folder on the server>] --name "<your name>" [--credit …] [--event …] [--city …] [--caption …]Your name, credit line and copyright on every picture the folder shows, and keywords added, without touching the captions, headlines or keywords each picture already has.
bun batch-iptc.ts "<your name>" ["<credit line>"] [keyword …]Shows only the picks with three stars or more, and exports them as JPEG to the folder you give, or beside the pictures.
bun export-picks.ts [<folder>]Scans the disks in the library and lists what exists more than once, whole folder copies first, with how much space they take. Give it a DNG and it also names the raw file it came from.
bun find-duplicates.ts [<picture.dng>]Two masks on the picture on screen: the top of the frame narrowed to its bright tones and darkened, and the skin tones lifted. Shows how a mask is put together from parts in a script.
bun local-adjustments.tsTakes the frames off an SFTP server as the camera sends them, reads the race numbers, looks them up in the start list, and delivers HD JPEGs with the riders' names in the metadata to a folder or Google Drive. Two races finishing on the same day are read in one pass.
bun race-delivery.ts <connection> <remote-path> <work-dir> <event-dirs> --drive <folder link>Every frame from the camera's SFTP is denoised once and delivered twice to Google Drive: cut to the player at HD, and whole at 4K for the desk to crop its own way.
bun football-delivery.ts <connection> <work-dir> <drive folder link> <match folder>For holding the shutter down: frames close in time are one burst, bursts that look alike are one moment, and only the best frame of each moment goes on, to a folder, Google Drive or the desk's FTP. The desk stops getting twenty near-identical frames.
bun burst-delivery.ts <connection> <remote-path> <work-dir> --to <connection>Shoot wide, deliver people: every runner is cut out of the frame in the same shape (9:16 unless you choose another), and so are pairs running together and groups, with the race number read off each runner.
bun runner-crops.ts <connection> <remote-path> <work-dir> --out <folder>Once per machine. Deliveries to Google Drive then renew their own access, instead of stopping an hour into a job.
bun drive-login.tsPrints every event the app sends as it happens: a picture rated, a folder opened, a delivery done. Where to start when writing a script of your own.
bun event-watcher.tsPrints one line per batch of finished pictures from a running workflow (what each step did, what went out), and a line for each error, start, stop and pause. Run it under Claude Code's Monitor, and the session looks at each batch and holds back or redoes a picture that is wrong.
bun follow-workflow.ts --batch-seconds 60Starts and stops the slideshow, steps through it, and rates and flags the picture on screen, one key at a time from a terminal: for showing pictures to a client while you sort them.
bun slideshow-controller.tsThese come with the app (Settings → Workflows → From Template…). A workflow someone sends you as a file goes in with Import… on the same page.
Ingest, extract JPEG, auto-crop 3:2, export JPEG 85%
Ingest → Use JPEG → Auto-Crop → Export
Ingest, denoise RAW, extract JPEG, auto-crop 4:5, export JPEG 92%
Ingest → Denoise → Use JPEG → Auto-Crop → Export
Ingest, extract JPEG, resize 2048px, export JPEG 80%
Ingest → Use JPEG → Resize → Export
Camera tether, denoise all, extract JPEG, export JPEG 95%
Camera Tether → Denoise → Use JPEG → Export
Ingest, find DNG, use RAW, auto-crop 3:2, export PNG
Ingest → Find DNG → Use Raw → Auto-Crop → Export
Watch an FTP folder, auto-crop 16:9 on the player, deliver HD JPEG to Google Drive
Ingest → Use Raw → Auto-Crop → Resize → Export
Watch an FTP folder, denoise with DxO, develop each picture with Auto, crop 3:2 (2:3 upright) to the subject the camera focused on and level it, add to an album published to eventphotos
Ingest → Denoise → Develop → Auto-Crop → Add to Album
The live event, with Claude looking at each picture before it goes into the album: one it fails is held back with its reason, or run again with the crop Claude proposes
Ingest → Denoise → Develop → Auto-Crop → Approve → Add to Album
Run on an event's folder of photographers: identify, resize 1920px, deliver JPEG 88% with names, captions and a .txt, split into with-numbers and without-numbers
Identify → Resize → Export
Run on an event's folder of photographers: identify, resize 3840px, deliver JPEG 88% with the numbers as keywords, split into with-numbers and without-numbers
Identify → Resize → Export
Ingest, extract JPEG, resize 3000px, auto-crop 16:9, export JPEG 88%
Ingest → Use JPEG → Auto-Crop → Resize → Export
Nothing shared yet.
A script you wrote, or a workflow saved with Settings → Workflows → Export… (it holds no password or key). It is read before it is published here.