[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-78692":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":9,"language":10,"languages":9,"totalLinesOfCode":9,"stars":11,"forks":12,"watchers":13,"openIssues":14,"contributorsCount":15,"subscribersCount":15,"size":15,"stars1d":16,"stars7d":17,"stars30d":18,"stars90d":15,"forks30d":15,"starsTrendScore":19,"compositeScore":20,"rankGlobal":9,"rankLanguage":9,"license":21,"archived":22,"fork":22,"defaultBranch":23,"hasWiki":24,"hasPages":22,"topics":25,"createdAt":9,"pushedAt":9,"updatedAt":26,"readmeContent":27,"aiSummary":28,"trendingCount":15,"starSnapshotCount":15,"syncStatus":29,"lastSyncTime":30,"discoverSource":31},78692,"autoresearch-genealogy","mattprusak\u002Fautoresearch-genealogy","mattprusak","Structured prompts, vault templates, and archive guides for AI-assisted genealogy research. Built for Claude Code.",null,"Ruby",1132,116,10,8,0,4,9,13,12,19.2,"MIT License",false,"main",true,[],"2026-06-12 02:03:48","# autoresearch-genealogy\n\nStructured prompts, vault templates, and research workflows for AI-assisted genealogy research. Built for Claude Code, adaptable to any AI tool or manual workflow.\n\nThis project extracts and generalizes methods developed during a real genealogy research effort that produced 105 files spanning 9 generations across 6 family lines, using Claude Code's autonomous research capabilities.\n\n## Who This Is For\n\n- **Genealogy researchers** who want to use AI to accelerate their family history work without sacrificing source rigor\n- **AI\u002Ftech enthusiasts** who want a concrete example of autonomous research loops applied to a humanities domain\n- **Anyone** who has a box of old photos, a DNA test, and unanswered questions about their family\n\n## Quick Start\n\nIf you are new to AI-assisted genealogy, start with [START_HERE.md](START_HERE.md). It routes you by what you already have: names, documents, DNA results, an existing tree, or a finding you need to verify.\n\n1. If you do not use Git, follow [Download And Start](guides\u002Fdownload-and-start.md).\n2. Copy the [vault-template](vault-template\u002F) folder into your Obsidian vault or any markdown editor.\n3. Fill in `Family_Tree.md` with what you already know. Mark living or possibly living people clearly and avoid exact birth dates for them.\n4. Use [Privacy Mode](guides\u002Fprivacy-mode.md) and [First Week Checklist](checklists\u002Ffirst-week-checklist.md).\n5. Use [Prompt Picker](guides\u002Fprompt-picker.md) to choose a verification or source-inventory prompt before expansion.\n6. Run [02 Cross-Reference Audit](prompts\u002F02-cross-reference-audit.md) or [05 Source Citation Audit](prompts\u002F05-source-citation-audit.md) before [01 Source-Backed Tree Expansion](prompts\u002F01-tree-expansion.md).\n\nSee [Getting Started](workflows\u002Fgetting-started.md) for the full walkthrough.\n\n## What's Included\n\n### Beginner Guides (`guides\u002F`)\n\nStart with [Download And Start](guides\u002Fdownload-and-start.md) if you do not use Git. Use [Prompt Picker](guides\u002Fprompt-picker.md) when you do not know what to run. The bundle guides package prompts into safer workflows:\n\n- [Beginner Pack](guides\u002Fbundles\u002Fbeginner-pack.md)\n- [Document Pack](guides\u002Fbundles\u002Fdocument-pack.md)\n- [DNA Pack](guides\u002Fbundles\u002Fdna-pack.md)\n- [Verification Pack](guides\u002Fbundles\u002Fverification-pack.md)\n- [Advanced Pack](guides\u002Fbundles\u002Fadvanced-pack.md)\n\nPrintable checklists cover [first-week setup](checklists\u002Ffirst-week-checklist.md), [scanning documents](checklists\u002Fscan-your-documents.md), [interviewing relatives](checklists\u002Finterview-a-relative.md), [verifying AI findings](checklists\u002Fverify-an-ai-finding.md), [adding ancestors](checklists\u002Fbefore-you-add-an-ancestor.md), and [sharing safely](checklists\u002Fshare-safely.md).\n\nFor a privacy-safe dry run, use the [First Run Walkthrough](walkthroughs\u002Ffirst-run.md) with the synthetic fixture.\n\nPlain-language reference pages explain [source grades](guides\u002Fplain-language\u002Fsource-grades.md), [evidence versus clues](guides\u002Fplain-language\u002Fevidence-vs-clues.md), [why AI can be wrong](guides\u002Fplain-language\u002Fwhy-ai-can-be-wrong.md), and [what counts as proof](guides\u002Fplain-language\u002Fwhat-counts-as-proof.md).\n\nIf you do not use Obsidian, follow [No-Obsidian Setup](guides\u002Fno-obsidian-setup.md). The template works as a normal folder of markdown files.\n\nBefore using public AI tools or sharing exports, follow [Privacy Mode](guides\u002Fprivacy-mode.md).\n\n### Prompts (`prompts\u002F`)\n\n12 autoresearch prompts designed for Claude Code's `\u002Fautoresearch` command. Each defines inputs to replace, a Goal, Metric, Direction, Verify condition, Guard rails, Iterations, and Protocol. They run autonomously: searching the web, updating your vault, and verifying their own work.\n\n| Prompt | Purpose |\n|---|---|\n| 01-tree-expansion | Review source-backed candidate relationships for deceased ancestors |\n| 02-cross-reference-audit | Find and fix discrepancies between your tree and source documents |\n| 03-findagrave-sweep | Locate Find a Grave memorials for every deceased ancestor |\n| 04-gedcom-completeness | Ensure your GEDCOM file matches your vault data |\n| 05-source-citation-audit | Verify every person file cites at least two independent sources |\n| 06-unresolved-persons | Identify and resolve unnamed people mentioned in your documents |\n| 07-timeline-gap-analysis | Find life events where records should exist but have not been found |\n| 08-open-question-resolution | Systematically attack every open research question |\n| 09-bygdebok-extraction | Extract data from digitized local history books (any country) |\n| 10-colonial-records-search | Search for colonial American ancestors in pre-1800 records |\n| 11-immigration-search | Locate passenger manifests and naturalization records |\n| 12-dna-chromosome-analysis | Analyze per-chromosome ancestry data to map genetic segments |\n\n### Vault Template (`vault-template\u002F`)\n\n19 files: a complete Obsidian vault starter kit with YAML frontmatter, plain markdown, readable anywhere.\n\n- **Core files**: Family tree, research log, open questions, data inventory, timeline, genetic profile, chromosome painting, witness network, unresolved persons, research strategy\n- **Templates**: Person, transcription, certificate, postcard, region, surname, hypothesis, draft letter\n\n### Archive Guides (`archives\u002F`)\n\n23 country and region-specific guides covering where to find records, what is free vs paid, and what AI tools can access directly vs what requires a browser.\n\n**Europe**: Ireland, England\u002FWales, Scotland, France, Italy, Spain\u002FPortugal, Germany, Netherlands, Austria, Hungary, Norway, Sweden, Poland, Russia\u002FUkraine\n\n**Americas**: USA (colonial, immigration, census, vital records), African American, Canada, Mexico\u002FLatin America\n\n**Oceania**: Australia\u002FNew Zealand\n\n**Cross-national**: Jewish genealogy\n\n### Reference Guides (`reference\u002F`)\n\n10 methodology documents: confidence tiers, source hierarchy, vault file manifest, DNA interpretation guardrails, naming conventions (patronymics, farm names, przydomki), GEDCOM format guide, common pitfalls, glossary, AI capabilities assessment, and the case for autoresearch in genealogy.\n\n### Workflows (`workflows\u002F`)\n\n7 step-by-step guides: getting started, OCR pipeline, new ancestor intake, document triage, oral history protocol, discrepancy resolution, phase planning.\n\n### Examples (`examples\u002F`)\n\n5 anonymized worked examples showing autoresearch in action: tree expansion session, cross-reference audit, DNA-to-genealogy mapping, name resolution, colonial deep dive.\n\n## Philosophy\n\n**Structured autonomous research with mechanical verification, not AI guessing.**\n\nGenealogy is different from most AI tasks. There is no compiler. Sources disagree with each other. Confidence is probabilistic, not binary. A name that appears as \"Sakkarias\" in one record and \"Zacharias\" in another might both be correct. A date listed as 1820 in one source and 1925 in another is almost certainly wrong somewhere.\n\nThe autoresearch approach adapts to this by:\n\n- **Defining measurable metrics** (count of sourced claims, count of resolved questions, count of remaining discrepancies)\n- **Requiring verification after every iteration** (cross-reference audit, not just accumulation)\n- **Logging negative results** (what you searched for and did not find is as important as what you found)\n- **Maintaining confidence tiers** (Strong Signal \u002F Moderate Signal \u002F Speculative) rather than treating all claims as equal\n- **Protecting living-person privacy** (autonomous prompts skip living people and redact exact private details)\n\nThis is inspired by Andrej Karpathy's autoresearch concept: autonomous goal-directed loops where the AI modifies, verifies, keeps or discards, and repeats. Applied to genealogy, the \"compiler\" is replaced by cross-referencing independent sources.\n\n## License\n\nMIT. See `LICENSE`.\n\n## Contributing\n\nContributions welcome. If you have prompts, workflows, or archive guides that worked for your research, open a PR. Please ensure all examples use placeholder names (no real family data).\n","autoresearch-genealogy 是一个用于辅助家谱研究的AI工具，提供了结构化的提示、模板和档案指南。该项目的核心功能包括结构化提示、保险库模板和研究工作流，这些都旨在利用AI（如Claude Code）来加速家谱研究过程，同时保持资料来源的严谨性。技术特点上，它使用Ruby语言编写，并且能够适应任何AI工具或手动工作流程。适合场景包括：希望借助AI加速家族历史研究的家谱研究人员；对自主研究循环在人文学科应用感兴趣的AI和技术爱好者；以及拥有旧照片、DNA测试结果并对家族历史有疑问的人士。通过提供详细的快速入门指南和分步操作说明，项目确保了即使是初学者也能轻松上手。",2,"2026-06-11 03:57:09","high_star"]