Developer Syntax Mode • Built for Cursor & Terminal

Speak Code at the Speed You Think.

Standard dictation mangles programming variables and terminal commands. Dev Syntax Mode automatically formats casing, flags, and code tokens on voice command.

The Speech-to-Code Problem

When you orchestrate autonomous coding agents (Claude Code, Cursor, Windsurf), you think in full architectural abstractions at 200 WPM, but your fingers can only type prompts at 50 WPM. Worse, ordinary speech engines format code like regular prose: kubectl becomes "cube cuddle", useEffect becomes "you sub fact", and --dry-run gets typed as "dash dash dry run" with broken spaces.

Dev Syntax Mode bridges this gap. It detects when your active window is an IDE or terminal, activating a specialized token grammar that converts voice cues into valid programming syntax in sub-450ms.

dev-syntax — active window: Cursor IDE / Terminal

# Voice input spoken: "const camel case user profile data equals await fetch user by id with flag force refresh"

const userProfileData = await fetchUserById({ forceRefresh: true });

✓ Automatic camelCase transformation • Argument object inferred • Zero spelling corrections needed

Voice Command Cheat Sheet

Speak naturally. The local engine recognizes standard programming casing conventions:

You Speak Dev Syntax Output Programming Category
camel case user account token userAccountToken JavaScript, TypeScript, Go
snake case audio buffer chunk audio_buffer_chunk Python, Rust, SQL
kebab case billing invoice item billing-invoice-item CSS, HTML, CLI arguments
screaming snake MAX RETRY LIMIT MAX_RETRY_LIMIT Constants & Configs
pascal case DatabaseConnectionPool DatabaseConnectionPool Classes, Types, Interfaces
flag recursive force -rf or --recursive --force Terminal / Shell CLI Flags
pipe to grep fatal error | grep "fatal error" Shell Pipelines

100% Confidential Code Base Guarantee

Unlike cloud dictation APIs that upload your proprietary variable names, database schemas, and API tokens to remote third-party servers, Speakeasy runs 100% on Apple Silicon hardware (Core ML + Rust). Your codebase never leaves your local machine.