Boost Prompt

Tools for improving and refining textual prompts directly inside VS Code.

Boost Prompt helps you turn a rough prompt into a clear, actionable, and well-formatted request for language models (Copilot Chat and compatible models).

Features

• Boost any selected text or whole file using a chosen language model.
• Registerable as a language-model tool (boostPrompt) for programmatic use from model-driven workflows.
• Persist and edit custom boosting instructions (stored in global storage as boost.prompt.md).
• Configurable file-patterns to enable the command only for the files you want.

Quick Start

1. Install the extension in VS Code.
2. Open a file that matches your configured patterns (default: *.prompt.md).
3. Select text (or leave selection empty to boost the whole file) and run the command Boost Prompt from the Command Palette or the editor context menu.

Commands

• Boost Prompt — Improve the active selection or whole document using the selected model.
• Boost Prompt: Select Model — Choose a language model to use (saved as preferred model).
• Boost Prompt: Edit Instructions — Open and edit the instruction template used to boost prompts.

Configuration

• boostPrompt.filePatterns (array of globs)
  • Default: ["*.prompt.md"]
  • Controls which files show the Boost Prompt command in the editor context and title menus. Use * or an empty array to enable for all files.
• boostPrompt.preferredModel (string)
  • Optional model name to use automatically. If unset, you'll be prompted to pick a model when boosting.

[!note] The extension discovers available Copilot-compatible chat models at activation. Make sure Copilot Chat is enabled in your environment.

Troubleshooting

• No models found: enable Copilot Chat in VS Code and try again.
• Boost not enabled for file: update boostPrompt.filePatterns to include your file extension (e.g. *.md, **/*.mdx).
• Check the BoostPrompt output channel for detailed logs and errors.

Extensibility

Boost Prompt exposes a languageModelTools contribution (boostPrompt) so other extensions or model-driven flows can reference the tool and pass promptText directly.

🙏 Acknowledgment

This extension, Boost Prompt, was inspired by the original idea from
Chris Dias – Prompt Boost.

While this project was built entirely from scratch and not derived from the original codebase, I want to give credit to Chris Dias for the concept that sparked the creation of this extension.