# Disco

Uncovers non-obvious statistical rules in tabular datasets without prior hypotheses. Useful when exploring complex feature interactions.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI · ★ 7
- Author: leap-laboratories
- License: MIT · policy=link-only · description only; the file stays with the author
- Source: https://github.com/leap-laboratories/discovery-engine
- Updated: 2026-10-07
- HTML: https://affpapa.org/en/nevibecode/mcp/disco

## What it does

Disco scans tabular records to detect combinations of attributes that noticeably alter a chosen outcome. Instead of prompting an LLM directly on raw CSV files, it trains machine learning models on a partition and extracts candidate rules using interpretability techniques.

The engine evaluates each discovered rule against a holdout test split and applies Benjamini-Hochberg FDR adjustments to control false discovery rates. Survived patterns undergo a semantic check against published papers to mark whether the finding is novel or already established.

Raw identifiers, surrogate columns, and metrics derived straight from the outcome must be dropped beforehand. Without this cleanup, the discovery pipeline will surface mathematically valid yet tautological relationships.

## Requirements
- Needs: Python and uv (commands run via uvx)
- Remote server: `https://disco.leap-labs.com/mcp`

## Install

### Claude Code (native platform)

**Platform CLI: remote server**
```bash
claude mcp add --transport http disco https://disco.leap-labs.com/mcp
```

### Codex

**Platform CLI: remote server**
```bash
codex mcp add disco --url https://disco.leap-labs.com/mcp
```

### Cursor

**Config file**
File `.cursor/mcp.json`:
```json
{
    "mcpServers": {
        "disco": {
            "url": "https://disco.leap-labs.com/mcp"
        }
    }
}
```

### GitHub Copilot

**Config file**
File `.vscode/mcp.json`:
```json
{
    "servers": {
        "disco": {
            "type": "http",
            "url": "https://disco.leap-labs.com/mcp"
        }
    }
}
```

### Gemini CLI

**Platform CLI: remote server**
```bash
gemini mcp add -t http disco https://disco.leap-labs.com/mcp
```

## When to use it
- detect multi-factor feature interactions across numeric and categorical columns
- confirm pattern reliability on unseen splits using false discovery rate corrections
- cross-reference detected rules with published academic research for novelty
- generate structured breakdowns containing condition thresholds and effect sizes

## Example prompts
- Search for novel patterns in `clinical_trials.csv` predicting the `recovery` outcome
- Run discovery on `sales_data.csv` to see what combinations drive `churn`

## Similar
- [World Monitor](https://affpapa.org/en/nevibecode/mcp/world-monitor.md) — MCP server · ★ 89609
- [Scrapling MCP Server](https://affpapa.org/en/nevibecode/mcp/scrapling-mcp-server.md) — MCP server · ★ 89899
- [MCP Toolbox for Databases](https://affpapa.org/en/nevibecode/mcp/googleapis-mcp-toolbox-for-databases.md) — MCP server · ★ 16730
- [PostHog MCP Server](https://affpapa.org/en/nevibecode/mcp/posthog-mcp-server.md) — MCP server · ★ 40459
- [dbx](https://affpapa.org/en/nevibecode/mcp/dbx.md) — MCP server · ★ 27844
- [Metabase](https://affpapa.org/en/nevibecode/mcp/metabase.md) — MCP server · ★ 49863

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md