mdmin 1.1.0

Rule-based markdown compression + context extraction for LLM consumption. Reduces token usage by 20-95%.

mdmin

Rule-based markdown compression for LLM consumption. Reduces token usage by 20–35%.

Website: mdmin.dev • npm: npmjs.com/package/mdmin

Install

pip install mdmin

Zero dependencies. Python 3.9+.

Usage

Python API

from mdmin import compress, estimate_tokens

result = compress(text, level="medium")

print(result.output)         # compressed text
print(result.stats.saved)    # tokens saved
print(result.stats.pct)      # % reduction
# CompressResult(output=..., stats=CompressionStats(input_tokens=2273, output_tokens=1765, saved=508, pct=22.3, ...))

CLI

# Compress a file (output to stdout)
mdmin compress README.md

# Save to file
mdmin compress README.md -o README.min.md

# Compression level
mdmin compress README.md --level aggressive

# Show token stats across all levels
mdmin stats README.md

# Pipe from stdin
cat file.md | mdmin compress -

Compression Levels

Level	Savings	What it does
light	~10%	Whitespace, comments, basic verbose patterns
medium	~20-25%	+ more verbose patterns, table compression, formatting cleanup
aggressive	~25-35%	+ article stripping, list compression, bold removal, dictionary dedup

What It Compresses

• Verbose phrases: 150+ patterns — "In order to" → "To", "Due to the fact that" → "Because"
• Whitespace: Blank lines, trailing spaces, decorative horizontal rules
• Tables: Markdown tables → compact CSV or key:value format
• Formatting: Redundant bold on headers, deep heading nesting, emphasis markers
• Lists: Short bullet lists → inline comma-separated (aggressive)
• Links: Empty titles, unused references, verbose alt text
• Dictionary dedup: Repeated phrases replaced with §1, §2 tokens

API Reference

compress(text: str, level: str = "medium") -> CompressResult

• level: "light" | "medium" | "aggressive"
• Returns CompressResult with .output (str) and .stats (CompressionStats)

estimate_tokens(text: str) -> int

Fast BPE token count estimate (no external dependencies).

CompressResult

result.output          # str — compressed text
result.stats           # CompressionStats

CompressionStats

stats.input_tokens     # int
stats.output_tokens    # int
stats.saved            # int (input - output)
stats.pct              # float (% saved)
stats.input_chars      # int
stats.output_chars     # int
stats.level            # str
stats.dictionary       # int (dedup entries created)

License

AGPL-3.0-only