Keep what the source said.
Read a supported RootsMagic snapshot without changing it. Keep original observations, file hashes and exact identity links in a separate local evidence store.
Evidence-first Genealogy · Free & open source
A name is a clue. Follow the evidence.
Keep the record, the competing explanations and your next useful question together. A family-history research toolkit built to leave room for uncertainty.
Read the example with no setup · Local tools · No account or AI subscription
Let uncertainty stay visible.
Separate a working conclusion from a plausible lead or a conflicting claim. Same names never trigger an automatic merge. Bad relationships stay available for review.
Give the next search a reason.
Compare sources, timelines, associates and hypotheses. Keep rejected ideas and negative searches. Record what is blocked and what evidence could change your mind.
See the trail, and the gaps.
Get a line audit, person dossier, hypothesis matrix, action queue and portable research state. Coverage reports keep their denominators and unknowns explicit.
Try it before installing
Two people. One name. A questionable link.
A workshop register names an apprentice called Ari Fable. A later index mixes up two people, and a copied parent claim does not fit the dates.
The complete invented case shows the evidence, the competing explanation, a rejected hypothesis and the next useful record to examine. You can read the mock sources yourself.
A working conclusion
Keep the question open.
The older Ari fits the stated age and named mother. The later index conflicts. Inspect the facing page before treating the identification as settled.
An invented teaching case. These labels organize judgment; they are not proof scores.
From a link to your own research
Start by opening a file.
The readable example works in a browser. The full research engine needs a desktop computer and Python 3.12 or newer, a free program that runs the toolkit.
- Read without installing. Download the offline example, extract the ZIP and double-click index.html.
- Run the complete workflow. Download the full source kit and extract it. Install Python if needed.
- Open the demo launcher. Double-click Start Demo.command on Mac or Start Demo.bat on Windows. It creates a fresh local database and opens the reports.
- Make a private research folder. Follow the guide to adapt the JSON examples and use the command-line tools. Your own records and reports stay separate from the public source.
Your research stays yours
Local tools. Deliberate decisions.
The runtime has no account, telemetry or automatic archive requests. It reads the files you choose and writes local databases and reports. The optional assistant skill helps organize research; running the toolkit uses no AI tokens.
This is a Python research toolkit with readable reports. Your own workflow currently uses editable JSON and terminal commands. There is no automatic ancestor discovery, archive crawler, DNA analysis, OCR engine or GEDCOM importer.
Your inputs and outputs can contain personal family information. Keep your working folder private and review anything you share. Privacy flags and source-rights labels are useful metadata, not anonymization.
Built for me. Yours to adapt.
A useful starting point.
I built this for me and my personal use. Make it your own, and feel free to improve mine. Hopefully it gives you a useful starting point, or at the very least some ideas. Cheers!
The full engine, design notes, templates, tests and optional research skill are on GitHub. This page's example comes from the pinned public release. The MIT license lets you use, change, share or sell your version, keeping the notice.