Skip to content
TakoAPI
Back to skills

data-cog

by nitishgargiitd

Coding Agents & IDEs
1.3kDownloads reported by clawskills.sh
5Stars
0Views
0 likesSign in to like this skill
—Not yet rated
Your rating

Brief

Your data has answers.

Description

What This Skill Does

CellCog is a data analysis agent that executes Python code server-side and returns results directly: charts, cleaned datasets, statistical reports, and interactive dashboards. Upload CSV, Excel, JSON, Parquet, or SQL exports and describe what you want to learn. It handles exploratory analysis, data cleaning, statistical testing, ML model training, and visualization without requiring you to run any code yourself.

Most AI tools return code and leave execution to you; CellCog runs the code and returns the actual charts, cleaned files, and conclusions.

When to Use It

  • Profiling an unfamiliar CSV to understand distributions, missing values, and outliers
  • Checking A/B test statistical significance and getting a ship-or-keep recommendation
  • Training a customer churn prediction model and seeing feature importance
  • Building an interactive HTML sales dashboard filterable by region and product
  • Cleaning mixed date formats, dollar-sign values, and duplicate rows in a dataset

Example Workflow

INPUT

User asks: Analyze this dataset and tell me everything interesting. [attaches customer_data.csv]

AGENT
  1. 1

    Profiles the dataset: row count, column types, missing values, and duplicates

  2. 2

    Computes summary statistics and key distributions for each column

  3. 3

    Identifies correlations and flags outliers or data quality issues

  4. 4

    Generates charts for the most significant findings

  5. 5

    Assembles results into an interactive HTML report with annotated insights

OUTPUT

Interactive HTML report with distribution charts, a correlation heatmap, flagged outliers, and 3-5 plain-English insights extracted from the data

Installation

clawhub install nitishgargiitd/data-cog

Run the CLI command, or paste the prompt into Claude Code, Cursor, Codex or whichever coding agent you use.

Links