aoc

name	aoc
description	Solve Advent of Code puzzles, algorithm challenges, and competitive programming problems. Activate when user provides AoC problem context/input, mentions solving puzzles/challenges, asks about algorithm selection (BFS, DFS, DP, etc.), or needs help with parsing structured input.

Advent of Code Solver

Language-agnostic problem-solving with TDD and correctness-first approach.

Workflow

1. READ      → Study problem + examples (examples are your spec)
2. PARSE     → Extract data structures from input
3. TEST      → Write tests from example input/output
4. IMPLEMENT → Minimal code to pass
5. RUN       → Execute on real input
6. ADAPT     → Refactor for Part 2

Solution Architecture

parse(input) → data structure
part1(data)  → answer
part2(data)  → answer

Parse once. Solve both parts. Test each function independently.

Algorithm Selection

Scenario	Algorithm
Unweighted shortest path	BFS
Path existence / exhaustive	DFS
Weighted shortest path	Dijkstra
Weighted + good heuristic	A*
"After N iterations..." (huge N)	Cycle detection
"Find minimum X such that..."	Binary search
"Count ways..." / "Min/max..."	Dynamic programming
Connected regions	Flood fill

Deep dive: See algorithms.md

Input Patterns

Format	Approach
Numbers in text	Regex -?\d+
Grid of chars	2D array or dict by coords
Blank-line groups	Split on \n\n first
Key-value pairs	Parse into map/dict
Instructions/opcodes	Pattern match each line

Grids: Use (row, col) with row↓. Sparse dict for infinite/sparse grids. Directions: UP=(-1,0), DOWN=(1,0), LEFT=(0,-1), RIGHT=(0,1)

Deep dive: See parsing.md

Part 2 Patterns

1. Scale up → Optimize algorithm
2. Add dimensions → 2D → 3D/4D
3. Many iterations → Find cycle, skip ahead
4. Reverse question → "Find X" → "Given X, find Y"
5. Add constraints → New rules or edge cases

Debugging

• Print intermediate state at each step
• Compare with example walkthrough
• Add assertions for every assumption
• Test parsing separately from logic
• Binary search on input size to isolate failures

Complexity Targets

Input Size	Target
n ≤ 20	O(2^n) OK
n ≤ 500	O(n³) OK
n ≤ 10,000	O(n²) OK
n ≤ 1,000,000	O(n log n)
n > 1,000,000	O(n) or O(log n)

Research Tools

# gh search code for algorithm implementations
gh search code "heapq.heappush" --language=python   # Dijkstra/priority queue
gh search code "collections.deque" --language=python # BFS patterns
gh search code "fn dijkstra" --language=rust

References

• algorithms.md - Graph traversal, DP, cycle detection, search
• parsing.md - Input formats, grids, coordinates, hex grids
• reference.md - Data structures, optimization, anti-patterns