Skip to content

About

Data-chat demo using Bauplan MCP, Atlassian Rovo MCP, GitHub and Claude Desktop

Resources

Stars

2 stars

Watchers

0 watching

Forks

Latest commit

 

History

8 Commits

Folders and files

Repository files navigation

Executive questions with MCP

A base project for answering executive questions about company data through a chat interface (for example Claude Code). A non-technical executive asks a question in natural language and gets back a concrete answer: a number, a table, or a chart. They never see the lakehouse, the pipelines, or the SQL underneath.

Every question resolves to one of two paths:

  1. The data already exists. A table answers the question, so the agent queries it and returns the result.
  2. The data does not exist yet. No table answers the question. The agent answers it with a query, returns the result, and files a Jira issue recording that query. Downstream automation converts it into a reviewed pipeline, and once the pull request merges the table is permanent.

The second path is what keeps answers consistent across executives over time: once a reviewed pipeline answers a question, the table it produces becomes the canonical source for that question.

Documentation

workflow.md is the spine. The others are references you reach for while following it.

Document When to read it
docs/workflow.md First. The end-to-end flow from a question to an answer: the query-versus-build decision, and how human review gates each new table.
docs/answering.md How to talk to the executive: answer directly, prefer charts, never invent an answer the data does not support. This is the doc to point a chat assistant's project instructions at.
docs/jira.md How to file the handoff Jira issue when no table answers a question: what the issue must contain and the instructions it carries to the implementing agent.
docs/semantics.md What the data means: the benchmark's business concepts, the metrics that matter, and the vocabulary to agree on before writing a query. Consult it whenever a question needs mapping to the data.
docs/lakehouse.md Where a new table belongs: the medallion architecture, bronze, silver, and gold layers, naming conventions, and the rules a pipeline must follow when you build one.

Operational rules for agents working in this repository (git workflow, Bauplan safety rules, uv usage, CLI versus SDK) live in CLAUDE.md.

About

Data-chat demo using Bauplan MCP, Atlassian Rovo MCP, GitHub and Claude Desktop

Resources

Stars

2 stars

Watchers

0 watching

Forks

Releases

Packages

Used by

Contributors