zapier code runs inside an invisible async function wrapper. this means top-level await and bare return statements are valid inside zapier, but invalid as standalone javascript. prettier tries to parse your code as a complete program and fails. codefmt handles this automatically by wrapping your code before formatting and stripping the wrapper afterward. it also detects zapier's newer sdk-in-code form (top-level `import` and `export default async function main(...)`) and parses it at module scope so the same prettier-style parse error doesn't happen there either.

code by zapier javascript injects inputData (your mapped input fields), output (assignable result), fetch (preinstalled, no require needed), callback for async completion, and console. the node standard library and fetch are available; the only npm package you can import is @zapier/zapier-sdk (enable it via the Packages panel in the code editor). codefmt recognizes all of these and won't flag them as undefined.

yes. zapier wraps your code in an async function, so top-level await works directly. you can also use return to send data to the next step, or call callback(err, output) for classic async-completion. codefmt preserves your async/await patterns during formatting.

not for general npm modules — lodash, axios, and moment are not available and cannot be required. the one exception is @zapier/zapier-sdk, which you enable via the Packages panel in the code editor; it lets you call your connected zapier app actions and make authenticated http requests through zapier. for anything else you either inline the logic or move to a different platform.

zapier python code injects input_data (mapped input fields), output (assignable result dict), StoreClient (class for zapier's key-value store: store = StoreClient('secret')), and requests (preinstalled http library). codefmt formats python code with ruff and recognizes all of these globals.