Pandas

claude mcp add --scope project clio-pandas -- clio-kit mcp-server pandas

Install the launcher first. Restart your client after setup. Configuration locations →

clio-kit doctor --server pandas --connect

Tools​

load_data

Load and parse data from CSV, Excel, JSON, Parquet, or HDF5 files with optional column selection and row limiting.

save_data

Save data to CSV, Excel, JSON, Parquet, or HDF5 with auto-detected format and optional index inclusion.

statistical_summary

Compute descriptive statistics, distribution analysis, and outlier detection for numerical and categorical columns.

correlation_analysis

Compute correlation matrices (Pearson, Spearman, or Kendall) with significance testing and strong-correlation detection.

hypothesis_testing

Run statistical hypothesis tests (t-test, chi-square, ANOVA, normality, Mann-Whitney) with test statistics and p-values; effect sizes are not computed.

handle_missing_data

Detect, impute, or remove missing values using strategies like mean/median/mode fill, forward/backward fill, or interpolation.

clean_data

Remove duplicates, detect outliers via IQR/Z-score, and optimize data types in a single pass. Writes <input>_cleaned.csv beside the input, overwriting any previous one.

groupby_operations

Group data by columns and apply aggregations with optional pre-filter. Writes output_file or <input>_grouped.csv. Existing output is refused unless overwrite=true; the input cannot be replaced.

merge_datasets

Join two datasets using inner, outer, left, or right joins on specified key columns. Writes <left_file>_merged.csv beside the left file, overwriting any previous one.

pivot_table

Create pivot tables with configurable row index, column headers, value columns, and aggregation function. Writes <input>_pivot.csv beside the input, overwriting any previous one.

time_series_operations

Resample, compute rolling statistics, create lag features, or difference a time series. Writes <input>_<operation>.csv beside the input, overwriting any previous one.

validate_data

Validate columns against rules for min/max range, data type, nullability, uniqueness, and regex patterns.

filter_data

Filter rows using comparison, membership, pattern-matching, and null-check operators across multiple columns. Writes <input>_filtered.csv beside the input (or output_file), overwriting any previous one.

optimize_memory

Analyze and reduce DataFrame memory usage through automatic dtype optimization and chunked-processing recommendations. Writes <input>_optimized.csv beside the input, overwriting any previous one.

profile_data

Generate a full dataset profile: shape, types, missing values, distributions, quality checks, and optional correlations.

profile_csv

Quickly profile a CSV file: row/column counts, per-column dtype, null counts, and min/max/mean for numeric columns.

Examples​

Replace example paths with your own files and check the returned results.

groupby_operations accepts output_file and defaults to <input>_grouped.csv. It refuses existing output unless overwrite=true is explicitly requested. The input file is never a valid output destination.

1. Data Loading and Profiling​

I have a large CSV file with sales data that I need to load and get a comprehensive profile including data types, missing values, and basic statistics.

Tools called:

• load_data - Load CSV file with intelligent format detection
• profile_data - Get comprehensive data profile and quality metrics
• statistical_summary - Generate descriptive statistics and distributions

2. Data Cleaning and Quality Assessment​

My dataset has missing values and outliers that need to be handled. I also want to remove duplicates and validate the data quality.

Tools called:

• handle_missing_data - Impute missing values with appropriate strategies
• clean_data - Remove outliers, duplicates, and optimize data types
• validate_data - Apply business rules and data quality checks

3. Statistical Analysis and Correlation​

Analyze the relationships between different variables in my dataset and perform hypothesis testing to validate my assumptions.

Tools called:

• correlation_analysis - Calculate correlation matrices with different methods
• hypothesis_testing - Perform t-tests, ANOVA, and normality tests
• statistical_summary - Generate comprehensive statistical insights

4. Data Transformation and Aggregation​

I need to group my sales data by region and product category, then create pivot tables for cross-analysis and merge with customer data.

Tools called:

• groupby_operations - Group data and perform multiple aggregations
• pivot_table - Create pivot tables with multi-level indexing
• merge_datasets - Join datasets using different merge strategies

5. Time Series Analysis and Filtering​

Analyze my time series data by resampling to different frequencies, calculating rolling averages, and filtering specific date ranges.

Tools called:

• time_series_operations - Resample, rolling windows, and lag features
• filter_data - Apply complex time-based filtering conditions
• statistical_summary - Analyze time series patterns and trends

For example, filter_conditions={"machine": {"eq": "gamma"}, "size": {"eq": 8000}} applies both conditions. Scalar equality and explicit operator/value pairs are also supported. Invalid conditions return an error without writing a filtered output file.

6. Data Export and Memory Optimization​

Optimize memory usage of my large dataset and export the cleaned data to multiple formats for different teams.

Tools called:

• optimize_memory - Reduce memory usage with dtype optimization
• save_data - Export to CSV, Excel, Parquet, and JSON formats
• profile_data - Verify optimization results and final data quality

Current calculation limits​

profile_csv uses a bounded retained sample, not necessarily all rows. Each file-based transformation reads its named input; use the returned output file for the next step. interpolate fills interior numeric gaps linearly by row position; [1, missing, 3, missing, 5] produces [1, 2, 3, 4, 5]. Endpoint, entirely missing and non-numeric gaps remain missing. Select the intended columns, check imputed_count, preserve the input and inspect the saved output. Irregular timestamps require a separately verified time-aware interpolation.