ToneGuard

ToneGuard — Thinks before you send. Ever hit Send and immediately regretted it? ToneGuard catches tone and clarity issues in your messages before they reach the other person. It only interrupts when it matters — no constant nagging for perfect grammar. How it works: - Write your message normally on Slack, Gmail, LinkedIn, or any site you add - When you hit Send, ToneGuard quickly checks your message with Claude AI - If the message looks good, it sends immediately — you won't even notice - If there's a tone or clarity issue, ToneGuard shows a suggestion with a clear explanation You decide: use the suggestion, edit it, or send your original message Features: - Real-time tone, clarity, and professionalism analysis - Smart suggestions with word-level diff highlighting - Explainable warnings showing what phrase was flagged and why - Rewrite modes for professional, warm, direct, de-escalating, boundary-setting, and concise messages - Voice preservation controls for lighter or stronger rewrites - Retry and diagnostic actions when analysis fails - Learns your writing style over time - Adjustable strictness (Gentle / Balanced / Strict) — globally or per site - Add your own custom rules in plain English - Works on Slack, Gmail, LinkedIn, by default + any site you add - Weekly stats dashboard showing your communication patterns 3-second undo after sending - Dark mode support - All data stays on your device — no tracking, no analytics - Local quality counters never store message text, prompts, recipients, API keys, emails, phone numbers, or URLs Privacy first: ToneGuard uses YOUR Claude API key — there's no middleman server. Your messages go directly from your browser to the Anthropic API and nowhere else. All settings and learning data are stored locally on your device. Requirements: - A Claude API key from Anthropic (https://console.anthropic.com) Chrome browser Open source: https://github.com/sumrae412/toneguard