Add reproducibility artifacts to remaining statistical skills

Author: mingzhangyang

skills/stat-assess-data-quality/SKILL.md

@@ -32,12 +32,14 @@ description: Assess data quality by reporting missing values, outliers, sample s
 6. Produce a missingness heatmap (rows = samples, columns = variables) if any missing values exist.
 7. Write a quality report (TSV) and missingness plot (PDF) to output directory.
 8. Summarize flags: list columns requiring attention and recommended actions (imputation, removal, transformation).
+9. Save the executable analysis script (`analysis_code.py` or `analysis_code.R`) and software version manifest (`session_info.txt`) used to produce results.
 
 ## Output Contract
 
 - Quality report table (TSV): column, dtype, n_non_missing, n_missing, missing_rate, n_outliers, outlier_rate, mean, sd, min, max, flags
 - Missingness heatmap (PDF, only if missing values present)
 - Summary of flagged columns (printed to stdout)
+- Reproducibility artifacts: `analysis_code.py` or `analysis_code.R`, plus `session_info.txt`
 
 ## Limits
 

skills/stat-bayesian-estimation/SKILL.md

@@ -37,6 +37,7 @@ description: Estimate model parameters using Bayesian inference (MCMC via Stan o
 8. Generate posterior predictive check plot (observed vs. simulated data).
 9. Write posterior summary (TSV), trace plots (PDF), and posterior plots (PDF) to output directory.
 10. Report convergence diagnostics; flag any parameters with R-hat ≥ 1.01 or ESS < 400.
+11. Save the executable analysis script (`analysis_code.py` or `analysis_code.R`) and software version manifest (`session_info.txt`) used to produce results.
 
 ## Output Contract
 
@@ -45,6 +46,7 @@ description: Estimate model parameters using Bayesian inference (MCMC via Stan o
 - Posterior distribution plots (PDF)
 - Posterior predictive check plot (PDF)
 - Convergence warnings (printed to stdout if any)
+- Reproducibility artifacts: `analysis_code.py` or `analysis_code.R`, plus `session_info.txt`
 
 ## Limits
 

skills/stat-cluster-samples/SKILL.md

@@ -40,6 +40,7 @@ description: Cluster samples or features using k-means or hierarchical clusterin
 4. Compute silhouette score for final cluster solution; report mean and per-sample scores.
 5. Assign and report cluster labels for each sample.
 6. Write cluster assignments (TSV), silhouette scores (TSV), heatmap or cluster plot (PDF) to output directory.
+7. Save the executable analysis script (`analysis_code.py` or `analysis_code.R`) and software version manifest (`session_info.txt`) used to produce results.
 
 ## Output Contract
 
@@ -48,6 +49,7 @@ description: Cluster samples or features using k-means or hierarchical clusterin
 - Clustered heatmap (PDF, hierarchical) or cluster scatter plot (PDF, k-means)
 - Elbow curve (PDF, k-means only when k is auto-selected)
 - Mean silhouette score (printed to stdout)
+- Reproducibility artifacts: `analysis_code.py` or `analysis_code.R`, plus `session_info.txt`
 
 ## Limits
 

skills/stat-compare-two-groups/SKILL.md

@@ -35,12 +35,14 @@ description: Compare a continuous variable between two groups with automatic sel
 7. Generate a box plot with individual data points and significance annotation.
 8. Write a structured result and plot (PDF) to output directory.
 9. Report method used, test statistic, p-value, effect size, confidence interval, and interpretation.
+10. Save the executable analysis script (`analysis_code.py` or `analysis_code.R`) and software version manifest (`session_info.txt`) used to produce results.
 
 ## Output Contract
 
 - Result (TSV): group1, group2, method_used, statistic, p_value, effect_size, effect_size_type, ci_lower, ci_upper, n_group1, n_group2
 - Box plot with data points (PDF)
 - Decision log explaining method selection (printed to stdout)
+- Reproducibility artifacts: `analysis_code.py` or `analysis_code.R`, plus `session_info.txt`
 
 ## Limits
 

skills/stat-fit-glm/SKILL.md

@@ -35,13 +35,15 @@ description: Fit a generalized linear model (Poisson, Binomial, or Gaussian fami
 6. Generate diagnostic plots: residuals vs. fitted (Pearson residuals), Q-Q plot of deviance residuals.
 7. Write results table (TSV), model summary (text), and diagnostic plots (PDF) to output directory.
 8. Report overdispersion finding and family choice justification.
+9. Save the executable analysis script (`analysis_code.py` or `analysis_code.R`) and software version manifest (`session_info.txt`) used to produce results.
 
 ## Output Contract
 
 - Coefficient table (TSV): term, estimate, std_error, z_statistic, p_value, exp_estimate, ci_lower, ci_upper
 - Model fit summary (TSV): family, link, null_deviance, residual_deviance, aic, df_residual
 - Diagnostic plots (PDF)
 - Overdispersion test result (Poisson only, printed to stdout)
+- Reproducibility artifacts: `analysis_code.py` or `analysis_code.R`, plus `session_info.txt`
 
 ## Limits
 

skills/stat-fit-linear-model/SKILL.md

@@ -33,13 +33,15 @@ description: Fit a linear regression model, report coefficients and model fit, a
 6. Generate residual diagnostic plots: (a) residuals vs. fitted, (b) Q-Q plot of residuals, (c) scale-location, (d) Cook's distance.
 7. Test regression assumptions: normality of residuals (Shapiro-Wilk), homoscedasticity (Breusch-Pagan), absence of influential points (Cook's distance > 4/n).
 8. Write results table (TSV), model summary (text), and diagnostic plots (PDF) to output directory.
+9. Save the executable analysis script (`analysis_code.py` or `analysis_code.R`) and software version manifest (`session_info.txt`) used to produce results.
 
 ## Output Contract
 
 - Coefficient table (TSV): term, estimate, std_error, t_statistic, p_value, ci_lower, ci_upper
 - Model fit summary (TSV): r_squared, adj_r_squared, f_statistic, f_p_value, aic, bic, n
 - Residual diagnostic plots (PDF, 4-panel)
 - Assumption test results (printed to stdout)
+- Reproducibility artifacts: `analysis_code.py` or `analysis_code.R`, plus `session_info.txt`
 
 ## Limits
 

skills/stat-learn-bayesian-network/SKILL.md

@@ -34,6 +34,7 @@ description: Learn the structure and conditional probabilities of a Bayesian net
 6. Visualize the learned DAG: nodes = variables, directed edges = learned dependencies, edge width proportional to bootstrap support.
 7. Perform a Markov blanket analysis for each variable (its direct causes, effects, and spouses).
 8. Write the DAG adjacency matrix (TSV), CPD parameters (TSV), network plot (PDF), and Markov blanket summary (TSV) to output directory.
+9. Save the executable analysis script (`analysis_code.py` or `analysis_code.R`) and software version manifest (`session_info.txt`) used to produce results.
 
 ## Output Contract
 
@@ -42,6 +43,7 @@ description: Learn the structure and conditional probabilities of a Bayesian net
 - Network visualization (PDF): DAG with node labels and edge confidence
 - Markov blanket summary (TSV): variable, parents, children, spouses
 - High-confidence edge list (TSV): from, to, bootstrap_support
+- Reproducibility artifacts: `analysis_code.py` or `analysis_code.R`, plus `session_info.txt`
 
 ## Limits
 

skills/stat-nonlinear-embedding/SKILL.md

@@ -36,12 +36,14 @@ description: Embed high-dimensional data into 2D using t-SNE or UMAP for explora
 6. If no metadata: generate plain scatter plot.
 7. Report: number of samples embedded, method and parameters used, runtime.
 8. Write embedding coordinates (TSV) and scatter plot (PDF) to output directory.
+9. Save the executable analysis script (`analysis_code.py` or `analysis_code.R`) and software version manifest (`session_info.txt`) used to produce results.
 
 ## Output Contract
 
 - Embedding coordinates (TSV): sample_id, dim1, dim2
 - Scatter plot (PDF): 2D embedding colored by metadata or uniform color
 - Parameters used (printed to stdout): method, all hyperparameters, random seed
+- Reproducibility artifacts: `analysis_code.py` or `analysis_code.R`, plus `session_info.txt`
 
 ## Limits
 

skills/stat-pairwise-correlation/SKILL.md

@@ -32,6 +32,7 @@ description: Compute pairwise correlations and significance tests across all num
 6. Generate a scatter plot matrix (pairs plot) for datasets with ≤ 10 variables.
 7. Report the top 20 strongest positive and top 20 strongest negative correlations.
 8. Write correlation matrix (TSV), adjusted p-value matrix (TSV), and plots (PDF) to output directory.
+9. Save the executable analysis script (`analysis_code.py` or `analysis_code.R`) and software version manifest (`session_info.txt`) used to produce results.
 
 ## Output Contract
 
@@ -40,6 +41,7 @@ description: Compute pairwise correlations and significance tests across all num
 - Correlation heatmap (PDF)
 - Pairs plot (PDF, only if ≤ 10 variables)
 - Top correlations summary table (TSV): var1, var2, correlation, p_value, adj_p_value
+- Reproducibility artifacts: `analysis_code.py` or `analysis_code.R`, plus `session_info.txt`
 
 ## Limits
 

skills/stat-pca/SKILL.md

@@ -35,6 +35,7 @@ description: Perform PCA on a numeric data matrix to reduce dimensionality, visu
 7. Plot PC1 vs. PC2 scatter; if metadata provided, color by specified annotation column; label outlier samples.
 8. Report top 20 features with highest absolute loading on PC1 and PC2.
 9. Write PC coordinates (TSV), loadings (TSV), variance explained (TSV), scree plot (PDF), and PCA scatter plot (PDF) to output directory.
+10. Save the executable analysis script (`analysis_code.py` or `analysis_code.R`) and software version manifest (`session_info.txt`) used to produce results.
 
 ## Output Contract
 
@@ -43,6 +44,7 @@ description: Perform PCA on a numeric data matrix to reduce dimensionality, visu
 - Variance explained table (TSV): PC, variance_explained, cumulative_variance
 - Scree plot (PDF)
 - PCA scatter plot (PDF, PC1 vs. PC2)
+- Reproducibility artifacts: `analysis_code.py` or `analysis_code.R`, plus `session_info.txt`
 
 ## Limits