Documentation

Prism Docs

29 tools across discovery, analysis, transformation, feature engineering, validation, and versioning.

Example workflow

  1. 1 Load a file with load_dataset("~/Downloads/sales.csv")
  2. 2 Run detect_data_quality_issues
  3. 3 Clean with handle_missing_values and remove_outliers
  4. 4 Encode categories and split with train_test_split
  5. 5 Export with save_processed_dataset

Tool reference

load_dataset
describe_dataset
detect_data_quality_issues
correlation_analysis
handle_missing_values
remove_outliers
encode_categorical_feature
train_test_split
normalize_categorical_text
harmonize_categorical_values
validate_action
save_processed_dataset

Configuration

Set MCP_DATA_DIR to control where datasets are saved. Default: ~/datasets