Read fields out of a screen snapshot
Turn the accessibility elements of one app's window ({ role, label }) into fields through that app's declarative parser row, with emails, card numbers and tokens removed before anything is returned
Turn the accessibility elements of one app's window ({ role, label }) into fields through that app's declarative parser row, with emails, card numbers and tokens removed before anything is returned
Give app (its name) and elements ([{ role, label }], as the Mac app's reader takes them). The first parser row whose app_contains matches the app is used, else the generic one; it selects elements by role and label pattern and cuts fields out with regular expressions. Every field is redacted with the redact patterns. Answers rows: [{ field: value }]. Triggers, throttle, parsers and redaction: read-data/snapshot-triggers.json; the governor that decides when to take a snapshot is createGovernor in the same file, used by the Mac app's reader.
Nothing: your AI starts it with your words.
Connect the AI you already use to Trillion once. Then ask in your own words; your AI finds this tool by what it does and runs it.
Say to your AI: “Use Trillion to turn the accessibility elements of one app's window ({ role, label }) into fields through that app's declarative parser row, with emails, card numbers and tokens removed before anything is returned”