databuddy-mcp

Query and mutate Databuddy analytics data through the MCP server tools.

1.1k|210|Updated Mar 11, 2025
One-click install
npx skills add https://github.com/databuddy-analytics/Databuddy --skill databuddy-mcp
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: databuddy-mcp
Source: https://github.com/databuddy-analytics/Databuddy/tree/main/.agents/skills/databuddy-mcp
Command: npx skills add https://github.com/databuddy-analytics/Databuddy --skill databuddy-mcp

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Working with the Databuddy MCP server requires knowing which tool to call, how to format dates and filters, and how to handle the durable investigation lifecycle. This Skill provides routing rules and conventions so analytics queries, error lookups, and investigation replies are executed correctly on the first attempt.

Core Features & Use Cases

  • Query Routing: Directs known-shape questions (top pages, recent errors, summary metrics) to get_data with batched queries, and discovery tasks to capabilities or get_schema.
  • Investigation Lifecycle: Covers listing investigations, retrieving evidence and history, replying with user context, and polling queued replies with the same replyId.
  • Conventions & Mutations: Standardizes website identification (ID, name, or domain), date presets versus from/to ranges, ClickHouse filter fields, and confirmation requirements for workspace mutations.
  • Use Case: A user asks why signups dropped last week. The Skill routes a comparison of the current and prior windows through get_data, then resumes the related investigation with reply_to_investigation to attach the findings.

Quick Start

Ask the assistant to use the Databuddy MCP server to show the top pages and recent errors for your website over the last 7 days.

Frequently Asked Questions about databuddy-mcp

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I query Databuddy analytics through the MCP server?

Use the get_data tool for known-shape questions like top pages, recent errors, or summary metrics, batching 2-10 queries with the queries array. For discovery of available data, call capabilities for the catalog or get_schema for column details.

How do I reply to a Databuddy investigation?

Call reply_to_investigation with the user context for the case, which resumes the same durable investigation. If the reply is queued or running, poll get_investigation and retry with the same replyId rather than creating a new one.

What date formats does the Databuddy MCP server accept?

Pass either a preset such as last_7d or both from and to dates in YYYY-MM-DD format. Never pass only one of from or to; if nothing is provided, the default window is last_7d.

Can I identify a Databuddy website by domain instead of ID?

Yes, any one of websiteId, websiteName, or websiteDomain works to identify a website in MCP tool calls. You only need to provide a single identifier, not all three.

When should I not use the databuddy-mcp skill?

Do not use it for SDK integration help, which belongs to the databuddy skill, or for monorepo implementation work, which belongs to databuddy-internal. It is scoped to querying and mutating data through the MCP server.