goalfydata

GoalfyData

Persist the user's business data as a structured dataset asset that is independent of projects and conversations and reusable and shareable across Agents / services / machines, and support developing datasets into publicly accessible data apps.

This document is the main guide with the complete execution flows. The sub-guides below provide supplementary reference:

• references/dataset-building-guide.md — business interview matrix, table naming rules, PG syntax pitfalls, pre-create checklist
• references/data-quality-guide.md — dirty-data classification and detection methods
• references/scheduled-sync-guide.md — script spec and fetch standard template, sandbox rules, external data-source templates (MySQL chunked, API paged), multi-table coordination, troubleshooting
• references/app-deploy-guide.md — app template structure, development conventions, version management details

The paths above are relative to the directory containing this SKILL.md, not to your working directory — references/ always sits next to this file. When reading a sub-guide or reopening this file, use the path the platform provided when loading the skill; if you cannot find it, search by filename (e.g. SKILL.md, dataset-building-guide*) instead of reconstructing path segments from memory. When installed as a plugin (Claude Code / Codex), note that the install path repeats "goalfydata" at several consecutive levels with a version directory in between (e.g. .../plugins/cache/goalfydata/goalfydata/<version>/skills/goalfydata/).

Prerequisites

Required MCP Server: goalfydata-mcp

The GoalfyData MCP Server provides 20 tools (15 dataset management-plane tools + 5 app development/deployment tools); all operations go through the GoalfyData backend API.

MCP configuration (streamable-http transport, API Key auth, Bearer scheme):

{
  "goalfydata-mcp": {
    "type": "streamable-http",
    "url": "https://mcp.goalfydata.ai/mcp",
    "headers": {
      "Authorization": "Bearer ${GOALFY_UDS_API_KEY}"
    }
  }
}