vrk plain

About

Strips markdown formatting and gives you plain prose. Headings, links, fences, and bullet markers are removed, but the actual text content is preserved. Useful before sending content to an LLM where formatting syntax wastes tokens.

The problem

You feed a markdown README to an LLM and waste tokens on formatting syntax - header markers, link URLs, fence markers, bullet characters. The content is the same but the token count is 20-30% higher than it needs to be.

Before and after

Before

# no standard CLI tool for this
python3 -c "
import re, sys
text = sys.stdin.read()
text = re.sub(r'#{1,6}\s+', '', text)
text = re.sub(r'\[([^\]]+)\]\([^\)]+\)', r'\1', text)
print(text)" < README.md

After

cat README.md | vrk plain

Example

cat README.md | vrk plain | vrk tok --budget 4000

Exit codes

Flags