This repository provides client-side modifications for the Cursor IDE to enhance the Claude 3.7 AI assistant. It targets Cursor users who want to increase Claude's context window, force its thinking level to "high," and customize its UI appearance without altering Anthropic's API.

How It Works

The project modifies specific JavaScript functions within Cursor's application files (workbench.desktop.main.js). It targets getEffectiveTokenLimit to increase the context window to 200,000 tokens, getModeThinkingLevel to ensure "high" level reasoning, and UI styling objects to customize model titles and subtitles. This approach leverages direct code manipulation within the client application.

Quick Start & Requirements

• Installation: Manual editing of workbench.desktop.main.js or use of provided Python script (cursor_claude_modifier.py) or batch files (run_modifier.bat, run_modifier.sh).
• Prerequisites: Cursor IDE installation. Python 3.x for the automated script.
• Resources: Requires locating and editing specific application files.
• Documentation: Comprehensive Guide to Claude 3.7 Customization in Cursor

Highlighted Details

• Increases Claude 3.7 token context window to 200,000.
• Forces Claude's thinking level to "high" for more detailed reasoning.
• Allows UI customization, including subtitles and styling (e.g., gradient text, pulsing effects).
• Includes an automated Python script for applying modifications and generating diffs.

Maintenance & Community

• The repository appears to be a personal project with no listed contributors, sponsorships, or community channels.
• Updates to Cursor may require reapplication of modifications.

Licensing & Compatibility

• The repository does not explicitly state a license.
• Modifications are client-side and do not affect API usage or costs. Compatibility is limited to the Cursor IDE.

Limitations & Caveats

Cursor updates will likely overwrite these client-side modifications, requiring reapplication. Some UI styling effects may not render consistently across different themes or Cursor versions. Increased thinking level may lead to higher token consumption.