solve-skeletonlisted
Install: claude install-skill Tenstu/Pass-LLM-with-LLM
# Solve Skeleton Skill
Bare-bones Python solve() skeletons for ACM/OJ problems. No logic inside — only structure,
I/O plumbing, and TODO markers. After filling TODOs, use the `algo-annotation` skill to add
Chinese comments and `# [防错]` markers.
## 1. Core Convention
### I/O
```python
input = sys.stdin.readline
n = int(input())
nums = list(map(int, input().split()))
```
For strings: `s = input().strip()`. For output: `print(ans)` or `print("\n".join(out))`.
**Never** use `sys.stdin.buffer`, `iter(data) + next(it)`, or `.buffer.read().split()`.
### Stage Separator
Five dashes, 58 equal signs. No trailing content.
```
# ============================================================
```
### 5-Phase Structure
Every solve() follows this layout:
```
def solve():
"""Input format: ...
Output format: ...
"""
input = sys.stdin.readline
# ============================================================
# Preprocess
# ============================================================
# ============================================================
# Algorithm
# ============================================================
# ============================================================
# Output
# ============================================================
if __name__ == "__main__":
solve()
```
The docstring must state input and output format — this is the contract with the grader.
## 2. Anti-Pattern Checklist
Never do these when wr