When you build a Coda integration that has to keep working, resolve column and table names to their IDs once and use IDs in every call. Names read better in examples but a user renaming a column in the UI breaks every name-based reference. List columns up front, cache the ID mapping, and fail loudly if a cached ID disappears.

Context: Official docs (Getting Started Guide, syncing a Google Calendar): documents that the examples use column names instead of IDs for simplicity, but warns that renaming a column makes scripts using its name start failing. The API reference echoes this: query resources by name or ID, but stick with IDs because names are fragile and doc users change them.