Disco
@leap-laboratories
Disco について
Discovery Engine — find novel, statistically validated patterns in tabular data
基本情報
設定
以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。
{
"mcpServers": {
"discovery-engine": {
"url": "https://disco.leap-labs.com/mcp",
"headers": {
"Authorization": "Bearer disco_..."
}
}
}
}ツール
ツールは検出されませんでした
ツールは README から自動的に抽出されます。メンテナーは ## Tools という見出しの下に記載することで、このタブに反映できます。
概要
What is Disco?
Disco is an MCP server and Python SDK that finds novel, statistically validated patterns in tabular data — feature interactions, subgroup effects, and conditional relationships — without requiring hypotheses. It validates each pattern on a hold‑out set, applies FDR correction, and checks results against academic literature.
How to use Disco?
Install with pip install discovery-engine-api, obtain an API key via the signup API or the developer dashboard, then create an Engine instance and call await engine.discover(file="data.csv", target_column="outcome"). For MCP, configure the server with "url": "https://disco.leap-labs.com/mcp" and set DISCOVERY_API_KEY as an environment variable.
Key features of Disco
- Unbiased, hypothesis‑free pattern discovery from tabular data
- Each pattern is validated on a hold‑out set with FDR correction
- Novelty classification against academic literature
- Structured output with conditions, effect sizes, p‑values, and citations
- Interactive web report for every analysis
- Public runs are free; private runs cost credits
Use cases of Disco
- Identify unknown feature interactions in clinical or biomedical datasets
- Discover subgroup effects in customer or operational data
- Find novel patterns in scientific research without pre‑defined hypotheses
- Understand which combinations of conditions drive a target outcome
FAQ from Disco
What data formats does Disco support?
CSV, TSV, Excel (.xlsx), JSON, Parquet, ARFF, and Feather. Maximum file size is 5 GB.
How much does Disco cost?
Public runs are free (results and data are published). Private runs cost credits: free tier gives 10 credits/month, Researcher $49/month (500 credits), Team $199/month (2000 credits), individual credits $0.10 each.
How is Disco different from standard data analysis tools?
Disco finds patterns that tools like pandas, AutoML, or LLMs miss. It does not start with a question but discovers statistically validated interactions and subgroup effects from the data itself.
Can I use Disco without an API key?
No. You must sign up via email (no password required) or the developer dashboard to get an API key. Public runs do not require a payment method.
How long does a typical analysis take?
A few minutes. The discover() method polls automatically and logs progress (queue position, ETA, pipeline step).
「その他」の他のコンテンツ
Awesome-MCP-ZH
yzflyMCP 资源精选， MCP指南，Claude MCP，MCP Servers, MCP Clients
MCP Registry
modelcontextprotocolA community driven registry service for Model Context Protocol (MCP) servers.
MCP Go 🚀
mark3labsA Go implementation of the Model Context Protocol (MCP), enabling seamless integration between LLM applications and external data sources and tools.
Blender
ahujasidOpen-source MCP to use Blender with any LLM
Reactive Resume
amruthpillaiA one-of-a-kind resume builder that keeps your privacy in mind. Completely secure, customizable, portable, open-source and free forever. Try it out today!
コメント