Family History Research Planning Skill
Forces a professional-genealogy research plan before any searching — solving the scattered, uncited, conflict-riddled approach amateur family historians typically fall into.
SKILL.md explicitly prohibits unsolicited web searches and requires information gathering, planning, and user confirmation before any research execution; README includes privacy best practices (CRAIGEN principles). No credential access, external writes, or destructive defaults. Deducted because the execution stage still relies on web_search and other external tools without systematic data-flow disclosure or rollback mechanisms.
Instructions are broadly self-consistent with clear progressive disclosure, but there is documentation drift: the References list in SKILL.md still cites `research-log-guidance.md` and `research-plan-guidance.md` as bare filenames although v1.0.5 moved them to references/, and the release workflow still packages under the old genealogy-research-skill name. No tests or edge cases; static review cannot exceed 10.
Trigger conditions (genealogy research requests, citations, evidence conflicts) are clearly described with rich examples; however the skill is entirely English- and US-centric (FamilySearch Wiki, LDSgenealogy, Ancestry), has no Chinese-language support, and its core resources have poor reachability from mainland China — deducted here and disclosed.
MIT license, semantic-versioned changelog, multiple release notes, and installation/troubleshooting docs show clear maintenance; deducted for path inconsistencies across docs, README's Version section saying 1.0 versus SKILL.md's 1.0.6, and License wording that does not match the actual MIT license.
Structured templates and professional workflows (GPS, Evidence Explained) with clearly defined, directly usable output formats; but all claims are static with no verified representative outputs or comparative-benefit evidence — capped at 7 statically and reduced to 6 for limited evidence.
Changelog, release workflow, and SHA256 checksum process exist, but there is no test suite, no execution evidence covering the skill's key paths, and no third-party corroboration; adherence to GPS/Evidence Explained is asserted only. Static cap of 5; scored 3.
- SKILL.md reference paths are inconsistent with the v1.0.5 directory structure (research-log-guidance.md / research-plan-guidance.md lack the references/ prefix) and may fail to load after install.
- The release package is still named genealogy-research-skill rather than family-history-planning, conflicting with installation docs.
- Core resources (Ancestry, FamilySearch, LDSgenealogy) target English/US users; mainland-China access is limited and there is no Chinese-language support.
- No tests or executed verification exist; output quality is independently unverified and template outputs should be reviewed by users.
- Version mismatch between README (1.0) and SKILL.md (1.0.6).
What does this skill do, and when should you use it?
A Claude Skill for family history and genealogy research, derived from the BMAD Method genealogy-assistant module. It enforces a plan-before-research workflow built on the Genealogical Proof Standard (GPS) and Evidence Explained citation methodology: it never performs unsolicited web searches, and instead gathers known information, clarifies objectives, and produces a structured research plan before any search — which happens only if the user explicitly requests execution. The skill provides four capabilities: research planning, citation generation (14+ source types), evidence conflict analysis, and research logging, backed by templates and reference documents. Version 1.0.6, MIT licensed, installable on both Claude.ai and Claude Code.
Loads SKILL.md as the core instruction file and pulls in reference docs on demand (citation templates, evidence evaluation, GPS guidelines, research strategies) from references/. It performs four operations: 1) Research planning — collects what the user already knows, breaks the goal into 3-7 answerable questions, lists record sources, builds a search strategy using the FAN principle (Family, Associates, Neighbors), and outputs a research plan document; 2) Citation creation — builds full reference notes, short forms, and bibliography entries per Evidence Explained standards for 14+ source types including census, vital, church, land, probate, military, immigration, newspaper, and database records; 3) Evidence analysis — classifies each source as original/derivative and primary/secondary, builds a comparison matrix, weights evidence to resolve conflicts, and outputs an evidence analysis report; 4) Research logging — documents sessions, positive and negative search results, and next steps. All outputs are generated as structured Markdown documents using templates in assets/templates/.
- A beginner who wants to research a great-grandfather but doesn't know where to start, and needs a concrete plan with prioritized sources
- A user whose census records say an ancestor was born in Ohio while the death certificate says Pennsylvania, and needs a systematic way to judge which is right
- A user who found a census record on Ancestry and needs a professionally formatted, reproducible citation
- An intermediate researcher who wants disciplined session logs to avoid re-searching databases already checked
- A family history writer who needs to verify a conclusion against all five GPS elements before publishing
What are this skill's strengths and limitations?
- Solid methodology: fully implements the five GPS elements and Evidence Explained citation standards, aligned with BCG professional norms
- The mandatory plan-before-search guardrail prevents the AI from burning tokens and context on aimless searching
- Progressive disclosure architecture: ~4k-word SKILL.md core with detailed references loaded on demand
- Five output templates and four reference documents produce deliverables usable for personal research or professional publication
- README includes explicit privacy guidance (no sensitive data about living people) and follows CRAIGEN responsible-AI principles
- MIT license, fully open and auditable
- Citation templates and location-resource guidance center on US records (federal/state census, NARA microfilm, county/state resources); researchers in other countries must adapt
- Strategy steps prioritize FamilySearch Wiki and LDSgenealogy.com, so resource coverage skews toward North America and English-language sources
- No automated test suite, and the README offers no evaluation data or user validation
- README install links reference a different repo name (genealogy-research-skill) than the source repository, which may cause download confusion
- Step 3 (research execution) depends on the host client providing a web_search tool; on clients without it, that step cannot be completed
How do you install this skill?
Download the latest family-history-planning-vX.X.X.zip from the repository's Releases page and extract it. Claude.ai users: enable Skills and Code execution under Settings > Capabilities, then upload the ZIP via Settings > Skills > Add Skill. Claude Code users: run mkdir -p ~/.claude/skills, then move the extracted family-history-planning folder into ~/.claude/skills/ and restart Claude Code. Note: some install links in the README point to emaynard/genealogy-research-skill rather than the source repo emaynard/claude-family-history-research-skill, so rely on the actual repository's Releases page for downloads.
How do you use this skill?
No manual invocation needed — Claude loads the skill automatically when you mention ancestor research, genealogy citations, or conflicting evidence. Typical prompts: "Help me plan research on my great-grandfather", "Create a citation for this census record", or "I have conflicting birth dates — help me figure out which is right." The skill will first ask you questions to gather known information and present a research plan; only after you explicitly say "please execute this research plan" will Claude use web search and other tools, documenting every search (including negative results) as it goes.