Browse / General-purpose / human-agentic-data-analysis

human-agentic-data-analysis

Vivid-Insights

Template workspace for collaborative data analysis with Claude Code

Open Vivid-Insights/human-agentic-data-analysis → Checked against the four rules.

Built for Claude Code

ships .claude CLAUDE.md and nothing for other agents.

In its own words

README.mdVivid-Insights/human-agentic-data-analysis

A folder for working through a data analysis with an agent, one step at a time, and coming out with a report you can defend.

It is an Obsidian vault with Claudian installed, so the conversation happens beside the documents it is writing. The agent's working contract is instructions.md, which it reads automatically.

What makes this different from just asking

Ask an agent to analyse a dataset and it will hand you a finished report in one pass: statistics run, charts drawn, conclusions written. It will look right. You will have no idea which choices were made on your behalf.

This folder is built to stop that. The agent does one step per request and then stops. It does not add analysis, findings or prose you did not ask for. It agrees the method with you before applying it, and it answers one sub-question at a time. The parts that require judgement stay with you, which is the whole point.

Getting started

  • Install the environment — Setup.md covers Obsidian, uv and Python.
  • Put your data in Attachments/.
  • Open a Claudian conversation and ask for step 1: import and describe the data.
  • Work through the eight steps, one request at a time.

If you want to see where this ends up before you start, read Sample Report/Happiness.md.

Quoting 30% of the README. Read the rest on GitHub ↗

Read from the repository's git tree on 2026-08-17.