lock in

A calm, cross-platform typing trainer — an app with a focus on typing without visual noise.

Current version: 0.5.0 · Changelog · GitHub

Screenshots

Features

• Per-character tracking — correct characters advance the caret, mistakes are marked and wait to be fixed, Backspace goes back.
• Live metrics — WPM, CPM/TPM, accuracy, and speed.
• Dictionaries — English and Russian word lists (with thematic topics: general, programming, tech), plus punctuation-rich sentences, quotes, and short passages.
• AI word-list generator — type any topic and generate a custom word list with your own LLM API key (any OpenAI-compatible endpoint: OpenAI, DeepSeek, Groq, …).
• Modes — echo (free typing), free, and timed (15 / 30 / 60 / 120 / 300 s).
• Shortcuts — Ctrl+Backspace deletes the word, Tab skips it.
• Session history — stored locally, with a Profile page and charts.
• Theming — 5 palettes × light/dark, UI and typing fonts, adjustable text size, error display modes, all in a style panel.
• Light / dark theme with system-preference support.
• Discord Rich Presence — shows the app as a Discord activity with the logo and the current mode/status.

Tech stack

• Frontend: React 18, TypeScript, Vite, CSS custom properties (design tokens).
• Desktop: Tauri 2 (Rust shell for the window, Discord presence, and the AI generation call).
• Tests: Vitest.

Project structure

src/
  components/     # TypingArea, TypingTest, EchoMode, Results, Profile, StylePanel, ChangelogModal, charts/
  hooks/          # useTypingSession, useStyle, useAiSettings, useAiWords
  lib/            # pure core + tests: engine, metrics, session, dictionary, history,
                  # echo, wordops, text, format, stats, style, changelog, ai, keys
  data/           # word lists and thematic texts (words/sentences/quotes/passages/topics)
src-tauri/        # Tauri 2 shell (Discord presence, AI generation command)
scripts/          # fetch-dictionaries.mjs, generate-icon.mjs

Getting started

Prerequisites

• Node.js ≥ 18 and pnpm.
• For the desktop build: Rust, and on Windows the MSVC C++ Build Tools (link.exe).

Development

pnpm install
pnpm dev        # frontend (Vite)
pnpm test       # unit tests
pnpm typecheck  # type check
pnpm build      # frontend build

Desktop

pnpm tauri dev     # run in dev mode
pnpm tauri build   # production build (NSIS .exe installer on Windows)

On Windows the Rust build requires MSVC C++ Build Tools:

winget install Microsoft.VisualStudio.2022.BuildTools

Metrics

Metric	Formula
WPM	(correct characters / 5) / minutes — 1 word = 5 characters
CPM / speed	correct characters / minutes
TPM	all keystrokes (correct + wrong) / minutes
Accuracy	correct / (correct + wrong), 0..1

The timer starts on the first keystroke (the Monkeytype convention).

Dictionaries

• Word lists — EN: google-10000-english (top 1500); RU: FrequencyWords / OpenSubtitles2018 (top 1500). Thematic topics (programming, tech) are hand-curated.
• Sentences, quotes, passages — curated public-domain and original prose with real punctuation, for EN and RU.

Word lists and thematic texts live in src/data/*.json; word lists regenerate with node scripts/fetch-dictionaries.mjs. See src/data/README.md for sources and licenses.

AI word generator

The Words mode has a Custom topic option: type a topic, hit Generate, and the app asks a cloud LLM (bring-your-own-key) for a list of words, sanitizes them, caches them locally, and feeds them into the test.

Configure it in the style panel (AI word generator section):

• API base URL — any OpenAI-compatible endpoint, e.g. https://api.deepseek.com or https://api.openai.com/v1.
• Model — e.g. deepseek-chat or gpt-4o-mini.
• API key — your own key, stored locally in the app config and sent only to the base URL above.

On the desktop build the request goes through Rust (reqwest); in the browser dev preview it uses a direct fetch (subject to the provider's CORS policy).

Theming

The whole UI is driven by design tokens (--bg, --card, --line, --typed, --accent, --err, fonts, and text size). Now 5 palettes are bundled — Dusk, Cocoa, Mocha Sage, Mushroom, and Espresso — each with light and dark themes. And planning adding more. The style panel also lets you pick UI/typing fonts, adjust the typing text size, and switch error styles.

Releasing

Releases are built automatically by GitHub Actions (.github/workflows/release.yml) when a v* tag is pushed.

1. Bump the version in package.json, src-tauri/tauri.conf.json, src-tauri/Cargo.toml, and src/lib/changelog.ts.

2. Add a new entry to CHANGELOG.md (Keep a Changelog).

3. Commit, then tag and push:

   git add -A && git commit -m "chore: release v0.5.0"
   git tag v0.5.0 && git push --tags

4. GitHub Actions builds installers for Windows/macOS/Linux and creates a draft release — review it and hit Publish.

To build manually instead: pnpm tauri build (the installer lands in src-tauri/target/release/bundle/).

License

MIT