Synthetic FSRS exercise for the Memly article: Anki FSRS Settings

This uses no real learner data and is not a personalized forecast. It is separate from Anki's built-in simulator.
Download these files into one empty folder:
https://memly.ai/downloads/anki-fsrs-settings/simulate.mjs
https://memly.ai/downloads/anki-fsrs-settings/package.json
https://memly.ai/downloads/anki-fsrs-settings/package-lock.json

With Node.js 20 or newer, run:
npm ci --ignore-scripts --no-audit --no-fund
TZ=UTC node simulate.mjs

The script writes simulation.json and summary.csv beside itself. It uses ts-fsrs 5.4.2 (MIT license), pinned by the lockfile. The full output includes all 21 default weights, seeds, assumptions and per-day counts. No production application dependency is changed.

1,000 new cards, 20 per day for 50 days; follow through day 119. Initial answer Good; later Good or Again sampled from the scheduler's own predicted recall. Five seeds; no missed days, review caps, Hard/Easy, fuzz or same-day steps. Later-review counts exclude the 1,000 initial study events. Seed ranges are not confidence intervals for people. Never describe these results as actual learning performance.