A simple script to enhance Claude 3.7 in Cursor by:
- Increasing token limit to 200,000 tokens
- Setting thinking level to "high" for more detailed reasoning
- Adding custom UI styling to distinguish the enhanced model
The script uses precise string matching to find and modify specific functions:
- Token limit: Uses multiple patterns and fallbacks to find
getEffectiveTokenLimit
function - Thinking level: Searches for
getModeThinkingLevel(e)
- UI styling: Searches for
_serializableTitle:()=>"claude-3.7-sonnet"}
This approach provides reliable matches across different Cursor versions, even when the code has been minified or slightly modified.
python hack_claude.py
The script automatically finds Cursor's workbench file and applies the modifications.
python hack_claude.py --token-mode all_models --ui-style red
Option | Values | Default | Description |
---|---|---|---|
--file , -f |
file path | (auto-detect) | Path to workbench.desktop.main.js |
--token-mode , -t |
claude37_only , all_models |
claude37_only |
Apply token limit to which models |
--ui-style , -u |
gradient , red , animated |
gradient |
UI style for the hacked model |
--skip-backup , -s |
flag | False | Skip backup creation |
Modify only Claude 3.7 with animated styling:
python hack_claude.py --ui-style animated
Apply the 200K limit to all models:
python hack_claude.py --token-mode all_models
- Creates backups before making changes
- Changes are lost when Cursor updates
- Client-side changes only - actual behavior still depends on Anthropic's API
- If you encounter issues, try using a backup file with
--file path/to/backup.js