CLIO Pandas

Community

Advanced pandas-based data analysis for LLMs

How to install ↓ Compare GitHub source repository ↗

Category: Filesystem #8 of 24
Stars: ★ 30 Popular
Transport: stdio (local process)
Runtime: Python 3.10+
Credentials: No credential needed
License: BSD-3-Clause
Last commit: 2 months ago
Tools: 16

60FMRS · C

This server provides a comprehensive set of pandas operations covering data loading, cleaning, statistical analysis, and optimization, with design aligned to real research needs, making it well-suited for AI-assisted data analysis. However, note its non-official status and local resource consumption; it is advisable to start with small-to-medium datasets and remain mindful of privacy and security.

Strongest · Maintenance 14/20 Weakest · Reliability 10/20

Reliability

10/20

Security and permissions

12/20

Maintenance

14/20

Documentation

12/20

Setup experience

12/20

Why each score

Reliability 10/20: Manifest declares 16 tools, stdio transport, and a PyPI package identifier 'clio-kit' with positional args 'mcp-server pandas' — internally coherent. However this is a static review with no committed tests or CI evidence specific to the pandas server's key paths (repo badges are unverified claims); tool-behavior fidelity cannot be confirmed. Version drift exists: manifest package version 2.10.6 vs README install 2.4.3, and README describes pandas as merely 'CSV data loading and filtering', inconsistent with the declared 16-tool breadth. Deducted for absent server-level test evidence and version inconsistency.
Security and permissions 12/20: DataFrame tools are local file operations with no declared network access or credential handling; main risks are file read/write side effects (save_data, clean_data) with no disclosed confirmation mechanism, path scoping, or data-flow disclosure in either manifest or README. No red-line issues found (no malware, credentials, or dangerous irreversible defaults), but incomplete permission/confirmation/scoping evidence caps the score at 12.
Maintenance 14/20: BSD-3-Clause license is clear; NSF funding, IIT research center ownership, Zulip community, and 39 open issues indicate active use and engagement. But version governance is muddled (2.10.6 vs 2.4.3 vs table contract versions) and no server-level security-response channel is evidenced — 'active but with governance/versioning gaps' tier.
Documentation 12/20: README covers installation, client configs (Cursor, Claude Code, VS Code, Claude Desktop), usage examples, and troubleshooting; however this is monorepo-level documentation with no pandas-specific tool parameter tables, limits, or cost notes, and README's description of pandas ('CSV data loading and filtering') contradicts the declared 16-tool capability, showing capability-doc drift.
Setup experience 12/20: One-command uv tool install plus JSON config examples for four mainstream clients with stdio transport make setup straightforward; but the install example version (2.4.3) mismatches the manifest package version (2.10.6), and without verifiable execution evidence of a working client connection, static calibration caps setup at 12.

Static review · not runListed 2026-09-08

Read the FMRS scoring method →

Fit and risk

What it can accessReads local filesWrites / deletes local filesRuns commands or code

Best for

Researchers and data scientists who want to perform complex data analysis through natural language
Rapid data exploration and cleaning in HPC environments without writing repetitive code
AI-assisted data preprocessing and feature engineering tasks

Not for

Large-scale analyses requiring real-time streaming or distributed processing
Tasks needing visual chart outputs (this server provides data operations only, no plotting)
Non-pandas users or those unfamiliar with Python data analysis

Required permissions

This server uses local file system read/write via stdio, so it can read and write CSV, Excel, JSON, and other files specified in AI conversations
Execution runs pandas/numpy code, consuming local CPU and memory resources

Risks and side effects

Loading extremely large files may cause memory exhaustion or slow client responses; consider using profile_csv to inspect data size first
User-provided data may contain sensitive information; file paths and contents are sent to the AI model, so privacy should be considered
As part of a research project, the server is not an official pandas component and may have edge cases not fully covered

Setup

Before you start

Runtime：Python 3.10+

Install Python 3.10+ and the uv package manager.
Run uv tool install 'clio-kit==2.10.6', then uv tool update-shell if prompted.
Add the JSON configuration example to your MCP client (e.g., Claude Desktop):

{
  "mcpServers": {
    "pandas-mcp": {
      "command": "clio-kit",
      "args": ["mcp-server", "pandas"]
    }
  }
}

Restart the client to start using it.

claude_desktop_config.json

{
  "mcpServers": {
    "pandas-mcp": {
      "command": "clio-kit",
      "args": ["mcp-server", "pandas"]
    }
  }
}

Shown for Claude Desktop. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json

{
  "servers": {
    "pandas-mcp": {
      "command": "clio-kit",
      "args": [
        "mcp-server",
        "pandas"
      ]
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

Terminal

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

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

Run clio-kit mcp-server pandas and confirm tools like load_data, statistical_summary, and profile_csv appear in your client's tool list; sending a profile_csv request proves the connection works.

Troubleshooting

If you get 'Server Not Found', run `clio-kit mcp-servers` to list available servers and ensure you use the name `pandas` not `pandas-mcp`
For import errors or missing dependencies, enter the `clio-kit-mcp-servers/pandas` directory and run `uv sync --all-extras --dev` then `uv run pandas-mcp`
If the uv or clio-kit command is not found, install uv following the official uv install guide, then run `uv tool install 'clio-kit==2.4.3'` and `uv tool update-shell`

Things to try

Once connected, you can ask your AI assistant things like:

Load sales_data.csv, clean missing values, and compute statistics grouped by region.
Generate a full profile of temperature.csv including missing values and distributions.
Merge data1.csv and data2.csv with an inner join on the id column and save as Parquet.
Run resampling and rolling statistics on monthly_revenue.csv.

Tools 16

load_data read-only

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

save_data writes

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

statistical_summary read-only

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

correlation_analysis read-only

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

hypothesis_testing read-only

Run statistical hypothesis tests (t-test, chi-square, ANOVA, normality, Mann-Whitney) with p-values and effect sizes.

handle_missing_data writes

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

clean_data writes

Remove duplicates, detect outliers via IQR/Z-score, and optimize data types in a single pass.

groupby_operations read-only

Group data by columns and apply aggregations (sum, mean, count, min, max, std, median) with optional pre-filter.

Show 8 more tools

merge_datasets read-only

Join two datasets using inner, outer, left, or right joins on specified key columns.

pivot_table read-only

Create pivot tables with configurable row index, column headers, value columns, and aggregation function.

time_series_operations read-only

Resample, compute rolling statistics, create lag features, or difference a time series.

validate_data read-only

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

filter_data read-only

Filter rows using comparison, membership, pattern-matching, and null-check operators across multiple columns.

optimize_memory read-only

Analyze and reduce DataFrame memory usage through automatic dtype optimization and chunked-processing recommendations.

profile_data read-only

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

profile_csv read-only

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

Use cases

Let an AI assistant load sales data, clean missing values, compute statistics by region, and save as a Parquet file.

Perform statistical summaries, correlation analysis, and hypothesis testing on DataFrames through natural language.

Automatically optimize memory usage of large DataFrames for more efficient processing.

Generate a full data profile to quickly understand data quality and structure.

Supported clients

Claude DesktopPartial support

Listed from the project's documentation, not tested by this site.

Overview

CLIO Pandas is an MCP server from CLIO Kit that wraps comprehensive pandas operations, enabling AI assistants to load data from CSV, Excel, JSON, Parquet, and HDF5 files, and perform statistical summaries, correlation analysis, hypothesis testing, missing data handling, data cleaning, grouping, merging, pivoting, time-series operations, filtering, validation, memory optimization, and data profiling through 16 specialized tools. Designed for scientific data analysis, it reduces repetitive coding across research workflows.

Similar servers

Filesystem MCP Server 84 · B

Secure local file read/write access

★ 91.4k · Tools 6 Compare with this →

Vault Cortex MCP Server 79 · B

Standalone MCP server for Obsidian vaults: hybrid search, notes & files, structured memory, tasks, OAuth 2.1.

★ 30 · Tools 30 Compare with this →

Obsidian Web MCP 68 · C

Secure remote access to your Obsidian vault over HTTPS with OAuth 2.0 authentication.

★ 185 · Tools 20 Compare with this →

Safe Docx MCP Server 68 · C

AI-native surgical editing of Word and OpenDocument files with formatting preservation

★ 46 Compare with this →

Compare with other MCP servers

More servers from this repository

Source revision 93e41d70d6c2 Data synced 2026-10-19 Read the FMRS scoring method