plugin.json

{
  "$schema": "https://agent-plugins.org/schemas/1.0.0/plugin.schema.json",
  "name": "datacleanist",
  "version": "1.0.0",
  "description": "Deterministic data-quality analysis (EDA) for tabular files: missing values, duplicates, outliers, correlations, VIF, mutual information, leakage checks, and an interactive report dashboard.",
  "author": {
    "name": "Mohd Anas Hussain",
    "email": "support@datacleanist.com",
    "url": "https://www.datacleanist.com"
  },
  "homepage": "https://www.datacleanist.com",
  "keywords": [
    "data quality",
    "EDA",
    "data cleaning",
    "exploratory data analysis",
    "machine learning"
  ],
  "extensions": {
    "com.openai": {
      "interface": {
        "displayName": "DataCleanist",
        "shortDescription": "EDA reports for your datasets",
        "longDescription": "DataCleanist analyzes tabular data files (CSV, TSV, Excel) and produces a deterministic exploratory data analysis: missing values, duplicates, outliers, distributions, correlations, VIF, mutual information, leakage suspects, PII flags, and a 0-100 data quality score with a five-dimension breakdown. Every number is computed by real code from the file itself - never estimated by a language model - and arrives with an interactive report card (Deep Dive statistics, per-column inspection, scatter explorer) rendered in the conversation.\n\nIntended users: data scientists, analysts, and ML engineers who want to understand and quality-check a dataset before modeling or cleaning it.\n\nLimitations: in-chat files up to 50 MB, 200,000 rows, and 300 columns; each user gets 5 analyses per rolling 30-day window; the tools analyze files - they do not clean, transform, export, or email data, and follow-up answers quote the computed summary rather than re-deriving statistics. Uploaded data is deleted automatically (raw files within 24 hours, derived outputs within 7 days) and is never used to train models. The full product - larger files, data cleaning, and chat - is available at datacleanist.com.",
        "developerName": "Mohd Anas Hussain",
        "category": "Productivity",
        "capabilities": [
          "Analyze datasets",
          "Generate EDA reports",
          "Inspect data quality"
        ],
        "websiteURL": "https://www.datacleanist.com",
        "supportURL": "https://www.datacleanist.com/about",
        "privacyPolicyURL": "https://www.datacleanist.com/privacy",
        "termsOfServiceURL": "https://www.datacleanist.com/terms",
        "defaultPrompt": [
          "Analyze this CSV and give me the full EDA report",
          "Which columns have missing values and can I trust them?",
          "Show me a demo of DataCleanist on a sample dataset"
        ],
        "brandColor": "#3B8EC2",
        "brandColorDark": "#60A5FA",
        "composerIcon": "./assets/icon.png",
        "logo": "./assets/logo.png"
      },
      "review": {
        "test_cases": {
          "positive": [
            {
              "description": "Full EDA on an attached dataset",
              "prompt": "Give me the full EDA report for the attached dataset.",
              "tools_triggered": "analyze_dataset, render_eda_report",
              "expected_behavior": "Reports the engine's computed results for the attached file - 891 rows x 12 columns, quality score 75/100, Cabin ~77% missing - and renders the interactive report card widget (Overview tab with quality breakdown and KPI tiles).",
              "file_attachment_urls": [
                "https://www.datacleanist.com/review/titanic_sample.csv"
              ]
            },
            {
              "description": "Demo sample analysis",
              "prompt": "Show me a demo of DataCleanist first.",
              "tools_triggered": "list_sample_datasets, analyze_sample_dataset, render_eda_report",
              "expected_behavior": "Lists the bundled Titanic sample, analyzes it with the deterministic engine, and renders the report card."
            },
            {
              "description": "Missing values and missingness mechanism",
              "prompt": "Which columns have missing values, and is that missingness random?",
              "tools_triggered": "analyze_dataset, render_eda_report",
              "expected_behavior": "Lists the columns with real missing counts (age, city, joined in the attached fixture) and quotes the computed MCAR/MAR verdicts from the Deep Dive summary - it never invents values.",
              "file_attachment_urls": [
                "https://www.datacleanist.com/review/messy_sample.csv"
              ]
            },
            {
              "description": "Correlations and outliers",
              "prompt": "What are the strongest correlations and the worst outliers in this dataset?",
              "tools_triggered": "analyze_dataset, render_eda_report",
              "expected_behavior": "Quotes real engine values (e.g. Pclass x Fare r = -0.55 on the Titanic sample) and the IQR-flagged outlier columns, pointing at the Relationships tab of the report card.",
              "file_attachment_urls": [
                "https://www.datacleanist.com/review/titanic_sample.csv"
              ]
            },
            {
              "description": "Deep-dive follow-up: normality",
              "prompt": "Is the Age column normally distributed, and how should I handle the missing ages?",
              "tools_triggered": "analyze_sample_dataset",
              "expected_behavior": "Answers from the computed Shapiro-Wilk verdict (not_normal) and the engine's recommended imputation (median) carried in the summary - no guessing or recomputation."
            }
          ],
          "negative": [
            {
              "description": "Oversized file is refused politely",
              "prompt": "Here is my 400 MB CSV - analyze it in chat for me.",
              "expected_behavior": "In-chat analysis supports up to 50 MB / 200,000 rows / 300 columns. Expected: the tool refuses with a clear limit message and points to datacleanist.com; it must not attempt or fabricate an analysis."
            },
            {
              "description": "No dataset attached",
              "prompt": "Analyze my dataset.",
              "expected_behavior": "No file is attached and no sample was requested. Expected: the plugin asks for the file (or offers the bundled demo) instead of inventing an analysis."
            },
            {
              "description": "Out-of-scope cleaning/export request",
              "prompt": "Clean my data and email me the cleaned CSV.",
              "expected_behavior": "The plugin performs analysis only - no cleaning, exports, or email. Expected: a clear refusal explaining what it can do here and that cleaning/export lives on datacleanist.com."
            }
          ]
        },
        "demo_recording_url": "https://www.datacleanist.com/review/demo.mp4",
        "commerce": false,
        "commerce_description": "This plugin does not sell products, subscriptions, or credits, and does not process payments. Analyses are free within the stated in-chat limits."
      },
      "publication": {
        "release_notes": "Initial release: deterministic EDA tools (analyze_dataset, analyze_sample_dataset, render_eda_report, list_sample_datasets) with an interactive report dashboard and deep-dive statistical findings."
      },
      "apps": "./.app.json"
    }
  }
}