Coding support for dyslexic developers, packaged as a VS Code extension.
Generic screen readers and dyslexia tools can't read code well: they stumble
over camelCase, skip operators, and drown you in raw compiler jargon.
DysLexCode understands code structure — it splits identifiers, speaks
operators as words, and turns error messages into short plain sentences.
Features (MVP)
Command
What it does
DysLexCode: Apply Readability Profile
One-command presets for font, line height, letter spacing, theme (Comfortable Spacing / OpenDyslexic / Low Contrast Light)
DysLexCode: Reset Readability Profile
Restores your original editor settings exactly
DysLexCode: Read Selection Aloud
Reads selected code (or the current line) with OS TTS — identifiers are split (getUserName → "get User Name") and operators spoken as words (=> → "arrow")
DysLexCode: Stop Reading
Stops speech
DysLexCode: Explain Errors in Plain Language
Sends diagnostics on the current line to a local LLM (Ollama, default) or Claude API and shows a short plain-language explanation, with optional read-aloud
Settings
dyslexcode.tts.rate — speech rate in wpm (default 180)
dyslexcode.tts.voice — TTS voice (macOS say -v name)
dyslexcode.llm.provider — ollama (default, fully local) / claude / off
dyslexcode.llm.ollamaUrl, dyslexcode.llm.ollamaModel — local Ollama endpoint & model
dyslexcode.llm.claudeModel — Claude model (default claude-opus-4-8)
dyslexcode.llm.language — explanation language (en / ja)
Claude API key
The key is never stored in settings. It is read from the ANTHROPIC_API_KEY
environment variable at runtime. Launch VS Code with the key injected, e.g.:
op run --env-file=.env -- code .
Development
npm install
npm run compile
# then F5 in VS Code (Run Extension)
Roadmap
[x] Readability profiles + settings UI
[x] Code read-aloud with identifier decomposition + OS TTS
[x] Diagnostics simplification via Ollama / Claude
[ ] Confusable-identifier detection (e.g. l vs I, rn vs m)