Family History Planning Skill
emaynard/claude-family-history-research-skillGenealogy skill that plans research before searching, then covers citations, conflicting evidence and logs under the GPS.
An evidence-first genealogy skill for the former USSR and Russian Empire: church and civil records, archives, evidence registers, an audit script and an HTML report.
Deep and carefully designed, but it is only days old, instructs the agent to make one CAPTCHA attempt, and can send emails and use your logged-in sessions when you allow it.
The author changed the skill files after our review (noticed Oct 29, 2026). The review describes the earlier version; we recheck changed skills weekly, the most significant changes first.
This single, very large skill turns an agent into a structured family-history researcher for the former Soviet republics and the Russian Empire. It starts with case intake (one question, known facts, who may be alive, what is off limits), then keeps registers of persons, artifacts, claims, relationships and a CSV search log. It resolves each event to the historical jurisdiction at the event date, builds name-variant matrices across Russian, Ukrainian, Polish, Romanian, German and Yiddish forms, and prefers original records over catalogs, indexes and finally trees and forums as leads.
Claims get named statuses (hypothesis, single-source, corroborated, contradicted and so on) instead of percentages, and copies of the same record share an independence group so they do not count as separate proof. An audit script checks the registers for broken references and weak corroboration before conclusions are written. The default output is an HTML report with each relationship laid out, contradictions and failed searches; GEDCOM export and PDF are optional.
The SKILL.md is about 54 KB and written in Russian, but the agent replies in the user's language. The folder has 40 reference files (government portals, correspondence, FamilySearch, browser handling), 13 scripts, 14 tests and templates, and the repo has about ten eval files. Optional workflows cover FamilySearch through a browser session or MCP connector, WWI records and drafting or sending archive requests when you authorize it.
A 54 KB SKILL.md with 40 reference files, 13 scripts, 14 tests, templates and an eval set, all in one skill.
Counted from the repository file tree. How we score depth.
A Russian-language genealogy research workflow with local Python helpers that keep case registers and query one public archive (gwar.mil.ru); the scripts read no credentials or cookies and write outside the skill folder. The instructions let the agent make one attempt at a real CAPTCHA in its browser without asking again, send archive requests by email once you have authorized that, and check the GitHub repo for skill updates (asking before updating). It processes personal data about named people.
The repository (78 files) on Oct 29, 2026. We had every file and read every script.
NVIDIA SkillSpector static scan: 75 pattern hits (analysis-evasion 35, MCP Rug Pull 18, Dangerous Code Execution 17, Data Exfiltration 3, Output Handling 1, MCP Least Privilege 1). The HIGH hits are in tests: os.environ is copied only to set PYTHONIOENCODING and subprocess.run executes the repo's own scripts. The remaining subprocess/MCP/'analysis-evasion' hits are test runs, a pinned-less 'npx skills' mention in translated READMEs, and a data.matricula-online.eu link in a reference document; no obfuscation or credential handling found.
A review, not a certification. Skills change after we check them — read the files before installing.
MIT — commercial use allowed.
Not really. It is a supervised research process with authorization steps for sends and site access, and it hands CAPTCHA, phone and payment steps back to you.
Genealogy skill that plans research before searching, then covers citations, conflicting evidence and logs under the GPS.
A 6-phase, 56-task OSINT workflow with a Python task runner, source grading, hypothesis analysis, templates and about 15 scripts.
It is an agent skill for genealogy in the former Soviet Union and Russian Empire. It builds evidence registers, resolves historical jurisdictions and name variants, grades claims, and produces an HTML report with sources.
Copy the repository's skill folder into your agent's skills directory (for example ~/.claude/skills/ or ~/.codex/skills/); it works with Codex and Claude Code. The agent replies in your language even though the instructions are in Russian.
The files are instructions and local helper scripts with an MIT license. It can browse archives and, with your authorization, send emails, and it tells the agent to try one CAPTCHA, so supervise it and decide what to permit.
This one focuses on former-USSR and Russian Empire archives with evidence registers and an audit script. The claude-family-history-research-skill is a general planning aid for genealogy, so it is lighter and not region specific.