Running in the browser

DuckPipe’s own source runs, completely unmodified, inside a browser tab via Pyodide — a real DuckDB engine, a real DAG, real fingerprint-based caching, no server involved at any point. ../examples/08_browser_wasm is a working page built around a concrete niche, not a generic demo: “is this file safe to send anywhere?” — checking an export for sensitive-looking columns (emails, phone numbers, SSNs, credit cards, IPs) without uploading it anywhere first, for anyone under an NDA, a legal hold, or a compliance policy where “upload it to a checker” is itself the violation. Pick “your own file” and the bytes go straight from your file picker into the sandbox with zero network requests — your data never leaves the tab — and reload the page after a run and it skips every task, because state persisted across the reload via IndexedDB, not just within one page’s JS lifetime. Both verified directly with an actual headless browser, not assumed. asyncio.run() inside the scheduler works unmodified because Pyodide uses WASM JSPI (stack switching), stable in Chrome 137+ today with no flag needed.

No remote sync or DuckLake backend in-browser yet — Pyodide’s DuckDB build has no runtime-loaded extensions, and whether fsspec works against Pyodide’s virtual filesystem is the next open spike, not claimed here. See the example’s own README for the full honest list, including how conservative the sensitive-column checks are.