GitHub radar
An AI skill that researches your ex-USSR family history, with proof
This is a skill for AI coding agents like Claude Code or Codex that turns them into careful, evidence-first genealogists for the former USSR and Russian Empire, marking clearly what is proven and what is only a guess.
The AI agent reads archival documents — church books, civil records, military and repression files — and separates what's in the document from its own assumptions.
When it helps
Useful once you already have documents or leads about an ancestor in the former USSR and want a rigorous analysis; it won't search the archives for you.
Pros
- Sorts findings into five clear statuses: hypothesis, single source, multiple, contradicted, unresolved
- Copies of one document never count as two independent sources
- Keeps people with the same name separate without clear proof
- Covers all 15 former Soviet republics using historical borders, not today's
Cons
- Won't search archives for you — that part is still your work
- Final PDF report needs Chrome, Chromium, or Edge installed locally
How to set it up — step by step
- 1Open a terminal on the computer where Claude Code or Codex is installed.
- 2Run git clone https://github.com/DenisSergeevitch/ex-ussr-kinship-research ~/.claude/skills/ex-ussr-kinship-research (for Codex, use ~/.codex/skills/ instead).
- 3Confirm Python 3 is already on your machine — no other software needed.
- 4Send the agent the text from the block below describing your case.
- 5Check which status the agent gave the finding and which sources it cited.
- 6Install Chrome, Chromium, or Edge if you want a finished PDF report.
Text for your agent
Copy this and send it to your agent — Claude Code, Codex, any of them:
Have your agent install the ex-ussr-kinship-research skill from https://github.com/DenisSergeevitch/ex-ussr-kinship-research and describe a concrete case - a great-grandfather exiled in the 1930s, or a 19th-century parish in Podolia - then show me what confirmation status the agent gave the finding and which sources it quoted.
Runs on a regular laptop - it's a set of instructions for the agent, no separate program needed, plain Python 3 is enough. The exception is the final PDF-report script, which needs Chrome, Chromium or Edge on the same machine.
Open on GitHub▌ More finds