Good skill

Skill WilliamWJHuang/agent-skill-evaluator/tests/fixtures/good-skill

Evaluate agent SKILL.md files for structure, security, quality, and domain correctness.

Install

npx -y skills add WilliamWJHuang/agent-skill-evaluator --skill good-skill

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

A well-structured statistical analysis skill that guides proper hypothesis testing with effect sizes and power analysis.

SKILL.md

2.9 KB, 565 tokens by cl100k_base, as published. Nobody here has run it

Statistical Analysis Guide

Overview

This skill guides the agent through proper statistical hypothesis testing, ensuring normality checks, appropriate test selection, effect size reporting, and multiple comparison corrections.

When to Use

Use this skill when the user wants to:

Compare two or more groups statistically
Test a hypothesis about differences or relationships
Determine if an observed effect is statistically significant

Decision Tree

IF comparing two groups:
  IF data is normal (Shapiro-Wilk p > 0.05) AND variances are equal:
    → Use independent samples t-test
  ELIF data is normal but variances unequal:
    → Use Welch's t-test
  ELSE (data is non-normal):
    → Use Mann-Whitney U test
    → Consider bootstrap confidence intervals

IF comparing 3+ groups:
  IF data is normal AND variances are homogeneous:
    → Use one-way ANOVA with post-hoc (Tukey HSD)
  ELSE:
    → Use Kruskal-Wallis test
    → Post-hoc: Dunn's test with Bonferroni correction

Guardrails

REFUSE to report only p-values. Always include effect size (Cohen's d, eta-squared, or odds ratio) and confidence intervals.
REFUSE to approve an experiment without power analysis showing power ≥ 80%.
WARN if multiple comparisons are performed without correction (Bonferroni, Benjamini-Hochberg, or Holm).
MUST NOT claim statistical significance without reporting the full context.

Output Format

Always report results in this structure:

Test: [test name]
Effect Size: [metric] = [value] ([interpretation])
95% CI: [lower, upper]
p-value: [value] (adjusted: [method])
Power: [value]
Assumptions checked: [list]

Edge Cases

Small samples (n < 30): Use exact tests or bootstrap methods
Tied data: Use appropriate tie-correction for rank-based tests
What if normality is borderline?: Report both parametric and non-parametric results

Common Mistakes

Using paired tests for independent samples (anti-pattern)
Reporting "trending toward significance" for p = 0.06 (do not use for this)
Cherry-picking seeds for reproducibility — report mean ± std across multiple seeds

Escape Hatch

Experienced users can override guardrails by explicitly stating: "I acknowledge [specific guardrail] and am proceeding because [justification]."

References

For detailed method guides, see the references/ directory.

Gives 0 of the 12 instructions most test skills give in 565 tokens

Counted across 964 of the 1,571 authors here whose files we hold, read 2026-08-07

close the browser when donein 55 of 964, across 12 files
wait for network idle statein 51 of 964, across 6 files
launch chromium in headless modein 49 of 964, across 6 files
use descriptive selectors for elementsin 49 of 964, across 6 files
run provided scripts with help flag firstin 49 of 964, across 6 files
add appropriate explicit waitsin 48 of 964, across 5 files
use bundled scripts as black boxesin 46 of 964, across 3 files
do not read script source codein 46 of 964, across 3 files
use sync playwright for scriptsin 46 of 964, across 3 files
inspect dom before executing actionsin 46 of 964, across 3 files
run the full test suitein 37 of 964
write the failing test firstin 29 of 964, across 23 files

Said here and by no other author read

use exact methods for small samples
check data normality and variance equality before testing
apply tie-corrections for rank-based tests on tied data
report both parametric and non-parametric results for borderline normality
report mean and standard deviation across multiple seeds
use the specified structure for reporting all results

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.