Codex-Assisted FamilySearch Workflows

A lightweight, public guide to using Codex, its in-app browser, and local-computer tools for careful family-history work on FamilySearch.org.

Scope and safety

• The repository tracks README.md, AGENTS.md, and the approved reusable skill in skills/familysearch-source-standardizer. It is not an application, research archive, or session log.
• Never store private genealogy, living-person data, screenshots, exports, credentials, cookies, or authentication artifacts here.
• Each session starts by reusing or opening FamilySearch and leaving it visible for Yuri to authenticate. Codex never handles credentials or authentication codes.
• Exploration is read-only by default. Any FamilySearch modification requires explicit approval.

Efficient browser workflow

• Known person: /en/tree/person/{tab}/{person-id}, where {tab} includes details, sources, memories, timeline, and related profile tabs.
• Known tree focus: /en/tree/pedigree/{view}/{person-id}. View Tree re-centers the last tree view.
• Unknown person: use Recents to identify the correct profile and branch before following relationships.
• During a task, keep a short, temporary map of active person IDs and verified parent, spouse, and child relationships. Recheck it before any modification and never persist it.
• Source review progresses from Sources list → source card → indexed record → Source Linker → accessible image or authorized external record. Stop wherever evidence or permission ends.
• FamilySearch loads dynamically. Wait for real content, then verify the person, heading, destination, and opened dialog before continuing. Leave reconnaissance forms through Cancel or Close, never Save.

Source-title convention

• Canonical: {STATUS} {EVENT} | {PRIMARY PERSON or PERSON A + PERSON B} | {UF - MUNICIPALITY} | {PROVENANCE}

🟡 means review has started; 🟢 means review has been completed using all evidence available through authorized routes. These statuses track progress, not historical reliability or freedom from human error.

Rules:

• Determine the event from the source and citation. Observed types include NASCIMENTO, BATISMO, CASAMENTO, ÓBITO, and MEMORIA PESSOAL.
• The person segment identifies the document's primary subject or subjects, independently of the profile where the source is attached. Birth and baptism are distinct; marriage uses both spouses.
• The canonical title never repeats the date. Verify and, with approval and supporting evidence, update Source Date to the primary event date. Preserve relevant secondary or administrative dates in the citation or notes.
• Use UF - MUNICIPALITY only when both parts are supported; never invent missing dates or places.
• Provenance follows the source's origin: use FS: Registro Civil for sources of civil origin and FS: Registro Paroquial for sources of parish origin, whether FamilySearch presents them through an index, OCR, transcription, or image; use Acervo familiar for material genuinely held by the family and Informante: {PERSON} for personal testimony. Do not copy a label from another title.
• Canonical spacing is one space after the emoji, around each |, and after :.
• A missing or restricted image does not prove completeness. Green is possible only after all information available through authorized routes has been processed.

What each source field does

• Source Date: the primary event date used to sort sources chronologically.
• Title: a recognizable identifier for the specific record; consistent wording also improves at-a-glance scanning and Source Box title searches.
• Citation: where the record can be found. Notes: transcription, explanation, and relevant secondary details.
• Reason to attach: why the source belongs to the person or relationship. Tags: the exact facts the source supports.

These interpretations follow FamilySearch's current guidance for creating sources, outside-source fields, source dates, primary people, and source tags.

Because Family Tree is shared, clear titles, citations, reasons, and tags help collaborators find the evidence, understand why it is attached, and evaluate conflicts. FamilySearch describes these collaboration benefits; the compact syntax itself is our convention, not an official FamilySearch standard.

Continuous improvement

After each substantive FamilySearch interaction, assess navigation and verification, review the guidance for safe compaction, and end the report with a short confirmation quiz about new conventions or unresolved assumptions. Change documentation or the approved skill only for validated, reusable, privacy-safe lessons, and keep its repository source synchronized with the local installation.

This independent project is not affiliated with or endorsed by FamilySearch.