claude-here 0.1.5

debug Python with claude.ai

claude-here

debug Python with claude.ai

usage

Just

pip install claude-here

then

import claude_here

and any uncaught exceptions will automatically launch a new conversation in claude.ai pre-filled with information for Claude to help you debug the error.

To debug something that isn't causing an exception, just add

breakpoint()

where you want to debug and it will automatically launch Claude with info about the current locals to help you debug.

nice things

Some extra features to note:

• If a breakpoint is hit multiple times, or an uncaught exception is raised after hitting a breakpoint, the information sent to Claude will be cumulative.
• If claude_here detects that webbrowser will use a text-based browser by default, it will instead print an OSC-8-compliant hyperlink to the terminal for you to open using a graphical browser. This is especially useful if you are debugging a remote machine.

breakpoint

breakpoint additionally supports some optional arguments:

• base_debugger determines the debugger to call in addition to launching Claude. Defaults to pdb.set_trace. Setting to None will set up a simple prompt to ask you when you're ready to continue execution.
• just_gather_info=True will gather information to send to Claude later (on an uncaught exception or breakpoint without just_gather_info=True) without actually launching Claude or the base_debugger.

environment variables

claude_here supports some additional configuration via environment variables:

• CLAUDE_HERE_PROJECT_ID can be set to a Claude Projects UUID (https://claude.ai/project/<this-part-here-is-the-uuid>) to launch your debugging conversations in.
• CLAUDE_HERE_MAX_CONTEXT_ITEMS can be set to an integer that will determine the maximum number of prior breakpoint contexts that will be sent to Claude.