Advent of Code 2025 (Gleam)

Includes parallel me- (mine/) and LLM- (llm/) generated implementations for side-by-side comparison. I'm using Crush and whatever models interest me. Each LLM-authored file includes a header indicating with model was the primary author.

Usage

$ gleam run                # Run all days with comparison (default)
$ gleam run -- <day>       # Run specific day with comparison
$ gleam run -- -e          # Run all days with example input
$ gleam run -- <day> -n    # Run specific day without comparison (mine/ only)
$ gleam run -- <day> -en   # Combine flags

Results

❯ gleam run
   Compiled in 0.04s
    Running aoc.main
=== Day 1 ===
Comparing implementations...

--- Part 1 ---
Mine: XXXX (5.3ms)
LLM:  XXXX (18.4ms)
✓ Results match
Mine 3.46x faster

--- Part 2 ---
Mine: XXXX (2.3ms)
LLM:  XXXX (17.3ms)
✓ Results match
Mine 7.50x faster