Optimize your AI prompts directly inside VS Code. PromptGuide analyzes prompt text for token usage, verbosity, and structural quality, helping you write premium, token-efficient instructions.
🚀 Features
Live Token Counter (Status Bar): Real-time token estimations for .md, .txt, and .prompt files. Compares original vs. optimized tokens so you can see your savings at a glance.
Inline Diagnostics & Warnings: Automatically flags common prompt issues (e.g., lack of role context, no output format constraint, action-less verbs, filler words).
One-Click Quick Fixes: Apply fixes instantly via the VS Code lightbulb menu (Ctrl+.).
Interactive Optimizer Panel: View detailed feedback, quality score breakdowns, and configure settings in a visual dashboard beside your editor.
Prompt Diff View: Inspect exact changes side-by-side in a native VS Code diff editor after optimizing.
Token Limit Mode: Monitor token usage against a selected limit (presets: 4K, 8K, 16K, 32K, or Custom). The status bar dynamically highlights in warning (amber) or error (red) when thresholds are crossed.
Quality Trend Chart: Analyze optimization progress per file with an interactive, SVG-based sparkline graph inside the panel.
Selection Hover breakdowns: Highlight any portion of prompt text and hover to see its estimated token count and input costs (GPT-4o vs Claude 3.5 Sonnet) over 1,000 runs.
Copilot Chat Participant: Ask @promptguide directly in GitHub Copilot Chat using /score, /optimize, or /tokens.
📦 Installation
To test or run PromptGuide locally:
Open the vscode-extension directory in VS Code.
Run npm install to install dependencies.
Run npm run build to compile the package.
Press F5 to open the Extension Development Host window.
⚙️ Settings
Customize PromptGuide's behavior:
promptguide.tokenModel: Set target tokenizer to Auto (Universal), GPT-4, GPT-3.5, Claude, or Llama/Mistral/Gemma.
promptguide.tokenBudget: Set a target token ceiling. The status bar colors change as you approach or exceed it.
promptguide.enableStatusBar: Toggle the status bar token counter.
