Prompt of the Day - Set Up a Personal Prompt Library
You can set up a personal prompt library by treating prompts like reusable research templates: categorize them by task, store them in one central place, and refine them over time as you test them in real projects.
1. Decide what your library is for
Think of your prompt library as a set of “forms” you fill in for recurring genealogy tasks rather than one‑off clever questions. For a working genealogist, this usually means:familyhistorystorytelling.wordpress+1
Research planning and locality questions
Record processing (census, civil, church, land, probate, pensions, immigration, Native American and Oklahoma Territory materials, etc.)
Evidence correlation and writing (proof discussions, narratives, blog posts, teaching materials)
DNA analysis support (match note organization, clustering, hypothesis drafting)
Start by listing 10–15 tasks you do often with AI; those become your first prompt categories.nextgengenealogy.substack+1
2. Choose where to store the prompts
Pick one “home” and stick to it so you actually use the library. For your stack, good options are:promptengineering+1
A Zotero group or collection called “AI Prompt Library,” with one note per prompt (tagged by record type and task).
A Markdown notebook (Obsidian, Better Notes, plain folders) with one file per category:
01_Research_Plans.md,02_Census_Analysis.md,03_DNA_Match_Notes.md, and so on.A Google Doc or spreadsheet if you want simple sharing and quick filtering (columns for name, goal, variables, tags, last updated).nextgengenealogy.substack+1
Since you already work heavily in Zotero and Markdown, a Zotero collection plus synced Markdown copies is ideal for fast copy‑paste into AI tools.
3. Use a consistent prompt template
Most prompt‑design guidance recommends a repeatable structure so every prompt is understandable “at a glance.” A simple, genealogy‑oriented template:aigenealogyinsights+2
Title – short and specific (for example, “Organize DNA match notes,” “Deed extraction table,” “Civil War pension timeline”).
When to use – 1–2 lines describing the scenario.
Inputs needed – what you must paste (for example, match list + notes; transcribed record).
Prompt text – the actual wording you’ll copy into the AI.
Variables to edit – placeholders like
[FOCUS ANCESTOR],[TIMEFRAME],[REPOSITORY].Tags – record type, task, and audience (for example,
#DNA #analysis #report,#census #correlation).promptengineering+1
This makes it easy to scan your library, tweak a few variables, and go.
4. Example library entry: organizing DNA match notes
Here’s a fully formed prompt you can drop straight into your library under a “DNA & match analysis” category.
Title: Organize DNA match notes into clusters and research priorities
When to use:
You have working notes on a group of DNA matches (for example, from Ancestry or MyHeritage) and want help organizing them by surname/locality clusters and priority, without AI drawing genealogical conclusions.
Inputs needed:
A list of matches with cM amounts and test site if known
Your free‑text notes for each match (how they might connect, surnames, locations, tree hints)
<prompt>
You are assisting a professional genealogist who uses DNA as one line of evidence alongside documentary records.
I will paste notes about several DNA matches who appear related to the same research focus. The notes may include: approximate cM amounts, test site, tree hints, hypothesized relationships, shared surnames, and shared localities.
Your tasks:
Normalize the notes into a table.
Columns: Match label or username, testing site (if stated), approximate shared DNA (cM), key surnames mentioned, key localities mentioned (with jurisdictions if given), hypothesized connection (in my words), and any documentary sources I mention.
Identify surname and locality clusters.
Group matches by recurring surnames and recurring localities.
For each cluster, list which matches belong to it and summarize what seems to connect them.
Suggest research priorities (without conclusions).
Based only on the notes I provide, suggest 3–5 matches or clusters that look most promising for further documentary research, and briefly explain why (for example, closer cM amount, overlapping locality with my focus ancestor, tree with relevant timeframe).
Do not state any relationship as proven. Use cautious language such as “may,” “possibly,” or “this cluster appears promising because…”.
Maintain an evidence‑focused, non‑speculative tone.
Do not fabricate trees or relatives. Work only with what my notes state.
Output format:
Section 1: “DNA matches table” as a table.
Section 2: “Surname and locality clusters.”
Section 3: “Suggested research priorities.”
When you are ready, ask me to paste my DNA match notes. Then wait for the notes before producing your answer.
</prompt>
You would save that exactly once, then duplicate/tweak for variants (for example, a version focused on one testing site or on a particular surname line).
5. Organize and tag by record type and task
A genealogist‑specific prompt library is most useful when you can say, “I’m doing [X] with [Y record type]” and immediately find a matching prompt. Consider categories like:familyhistorystorytelling.wordpress+2
Research design & context – research plans, locality summaries, timeline scaffolds, negative search logs.
Record‑specific prompts – census, civil registration, church registers, land/deeds, probate, pensions, military, immigration/naturalization, Native American/Oklahoma Territory records, city directories, newspapers.
Analysis & correlation – correlation tables, conflict lists, FAN summaries, hypothesis articulation.
DNA & FAN – match note organizers, cluster summaries, segment‑level question templates.
Writing & teaching – ancestor sketches, blog post scaffolds, report sections, lesson plans, handouts.ngsgenealogy+2
Within each category, tag prompts by difficulty (basic/advanced), evidence phase (discovery/analysis/writing), and audience (client, blog readers, students).
6. Start small, then refine and version
Prompt‑engineering advice and AI‑for‑genealogy guides stress that libraries are living documents you improve after real use, not one‑time projects.ngsgenealogy+3
Begin with 8–12 “starter” prompts tied to your most common tasks (for example, research planning, pension timelines, deed extraction, DNA match notes, ancestor narratives).
As you test, add “Usage notes” under each prompt: what worked, where it failed, what you changed.
Keep a simple version marker in the title (for example, “v1.2”) when you substantially revise a prompt so you can roll back if a later edit is less effective.reddit+1
After a month or two, you will have a compact, highly tailored library that mirrors your personal methodology and vocabulary.
Comments
Post a Comment