Align local benchmark skill levels with CI

[luccabb]

Summary

• Changed local benchmark default skill levels from 1 2 3 4 5 to 3 4 5 to match CI configuration
• Added scripts/** to benchmark CI path trigger so script changes trigger the workflow

Test plan

• Verify CI benchmark workflow triggers on this PR (since scripts/** path is now included)
• Compare local and CI benchmark results for skill levels 3, 4, 5

[greptile-apps]

Greptile Overview

Greptile Summary

This PR makes local benchmarking match the CI benchmark configuration by narrowing the default Stockfish skill levels in scripts/benchmark.sh to 3 4 5. It also updates the benchmark GitHub Actions workflow path filter so changes under scripts/** will trigger the benchmark job, ensuring the benchmark script and CI workflow stay in sync.

I didn’t find any issues that need fixing before merge in the touched files; both changes are minimal and consistent with the stated intent.

Confidence Score: 5/5

• This PR is safe to merge with minimal risk.
• The PR only adjusts benchmark configuration: adding scripts/** to a workflow path filter and changing the benchmark script’s default skill levels while preserving env overrides. No production code paths are affected, and the changes are syntactically valid.
• No files require special attention

Important Files Changed

Filename	Overview
.github/workflows/benchmark.yml	Adds 'scripts/**' to the workflow path filter so benchmark CI runs when benchmark scripts change; change is syntactically valid and matches intent.
scripts/benchmark.sh	Changes default SKILL_LEVELS from '1 2 3 4 5' to '3 4 5' to align with CI; keeps env override behavior intact.

Sequence Diagram

sequenceDiagram
    participant Dev as Developer
    participant Git as GitHub
    participant GA as GitHub Actions
    participant Script as scripts/benchmark.sh

    Dev->>Git: Push changes to PR
    Git-->>GA: Trigger workflow (paths include moonfish/**, opening_book/**, scripts/**, pyproject.toml, requirements.txt)
    GA->>Script: Run benchmark script
    Script-->>GA: Uses default SKILL_LEVELS="3 4 5" (unless overridden by env)
    GA-->>Git: Publish benchmark job result

Loading

[greptile-apps]

_{2 files reviewed, no comments}

_{Edit Code Review Agent Settings | Greptile}

[github-actions]

🔬 Stockfish Benchmark Results

vs Stockfish Skill Level 3

Metric	Wins	Losses	Draws	Total	Win %
Overall	43	47	10	100	43.0%
As White	22	23	5	50	44.0%
As Black	21	24	5	50	42.0%

Non-checkmate endings:

• Draw by 3-fold repetition: 9
• Draw by fifty moves rule: 1

vs Stockfish Skill Level 4

Metric	Wins	Losses	Draws	Total	Win %
Overall	16	75	9	100	16.0%
As White	8	39	3	50	16.0%
As Black	8	36	6	50	16.0%

Non-checkmate endings:

• Draw by 3-fold repetition: 9

vs Stockfish Skill Level 5

Metric	Wins	Losses	Draws	Total	Win %
Overall	12	80	8	100	12.0%
As White	5	40	5	50	10.0%
As Black	7	40	3	50	14.0%

Non-checkmate endings:

• Draw by 3-fold repetition: 6

Configuration

• 5 chunks × 20 rounds × 3 skill levels = 300 total games
• Each opening played with colors reversed (-repeat) for fairness
• Moonfish: 60s per move
• Stockfish: 60+5 time control