Experimental tool for concatenating and formatting files under the current directory into an LLM prompt, respecting .gitignore and .promptignore files by default.

When asking a chat-based LLM like ChatGPT o1 something about code, you want to provide as much context as possible - often the limiting factor is the size of the context window. I'm using this tool as something I can easily add functionality to as needed, e.g. being able to count tokens for individual files in order to be able to work out which files to trim from a prompt, or being able to easily ignore certain file types. For a more mature tool, see https://github.com/mufeedvh/code2prompt.

prompt # copies straight to clipboard and prints summary prompt --format json --stdout # prints prompt content as json to stdout prompt -p src/ app/ -e out/ # include/exclude certain paths/globs prompt --no-gitignore # include files that are normally skipped by gitignore

# Images *.png *.jpg *.jpeg *.gif *.bmp *.tiff *.ico *.webp *.svgz # Audio *.mp3 *.wav *.ogg *.flac # Video *.mp4 *.mkv *.mov *.avi *.wmv # Archives / Compressed files *.zip *.tar* *.gz *.bz2 *.7z *.rar *.tgz # Documents (binary/non-text) *.pdf *.doc *.docx *.xls *.xlsx *.ppt *.pptx # Executables *.exe *.dll *.so *.dylib *.bin *.dat *.wasm # Fonts *.ttf *.otf *.woff *.woff2 # Lockfiles yarn.lock package-lock.json pnpm-lock.yaml bun.lockb go.sum Cargo.lock # Secret/test files *.crt *.pem *.crl *.ski *.der # Minified *.min.js