LLM-powered ww help | Generated by AI
Done. Here’s what was implemented:
New command: ww help <file_path>
It uses the LLM (deepseek/deepseek-v4-flash) to read a module file, look up how it’s wired in main.py, and tell you the exact ww command to use.
ww help ww/gcp_speech/transcribe.py
ww help ww/ffmpeg/merge.py
ww help ~/projects/ww/ww/display/appearance.py
Files changed:
ww/help_llm/help.py(new) — the help module: reads the target file + scansmain.pyfor its import/wiring, sends both to the LLM, prints the explanationww/help_llm/__init__.py(new) — empty package initww/main.py— addedhelpas a separate command group (ww help <path>→ LLM help,ww help→ existing help;--help/-hstill shows the help page unchanged)~/.zsh/completions/_ww— addedhelpto commands list and dispatch case (tab-completes file paths)
