AI Prompts & Tools for Genealogists
Over the past years, I’ve developed a growing collection of AI prompts designed specifically for genealogists and family historians. These aren’t generic ChatGPT tricks—they’re carefully tested tools for real research tasks: transcribing handwritten documents, restoring damaged photographs, extracting facts from messy sources, and more.
Everything here is open source and free to use under a Creative Commons license. You can copy any prompt into your preferred AI tool (ChatGPT, Claude, Gemini, etc.) and start using it immediately.
Full repository: Open-Genealogy on GitHub
Quick Start: Recommended Prompts
If you’re new to using AI prompts for genealogy, start with these. They represent the most refined, reliable versions in each category:
| Task | Recommended Prompt |
|---|---|
| Transcribing handwritten documents | Steve’s OCR-HTR Transcription Tool v08 |
| Restoring historical photographs | Historical Photograph Restoration v2 |
| Deep research workflows | Research Agent Assignment v2.1 |
| GPS-compliant genealogy assistant | Contract-First Genealogy System Prompt v3.1 |
Transcription & Handwriting Recognition (HTR)
These prompts help you transcribe historical documents—from faded census pages to your great-grandmother’s letters. They emphasize diplomatic transcription: preserving exactly what’s on the page, including spelling, punctuation, and layout, while clearly marking uncertain readings.
- Steve’s OCR-HTR Transcription Tool v08 (recommended)
The most comprehensive version. Merges diplomatic fidelity with JSON metadata, multi-phase verification, and unified notation standards. Use this unless you have a specific reason not to. - Steve’s Transcription Tool v07
Optimized for reliable uncertainty tagging and stable structured outputs. A solid alternative if v08 feels like overkill for your task. - Unofficial HTR Transcription (Humphries Method) v1
A minimalist approach focused on strict verbatim capture. Use when you want a shorter, tighter “just transcribe” workflow. - Describe → Transcribe → Analyze → Interpret
A compact four-step scaffold: first describe the artifact, then transcribe verbatim, analyze its content, and interpret meaning. Useful as a quick workflow reminder.
Legacy versions (v02–v06.5) are available in the GitHub repository for reference.
Historical Photo Restoration
AI image generators have become remarkably capable at repairing damaged photographs. These prompts guide the process with historical accuracy in mind—preserving period-appropriate details, avoiding anachronisms, and respecting the evidence.
- Historical Photograph Restoration v2 (recommended)
A universal, high-standard restoration prompt that adapts to any subject count, damage severity, and photographic era. Start here. - Steve’s Photo Damage Removal Specialist v3
Intensive museum-grade restoration. Begins with detailed analysis of subjects, context, period, and damage type, then removes damage comprehensively while avoiding AI artifacts. - Steve’s Historic Photo Reconstructor v3
For severely damaged photos where significant reconstruction is needed. Directs the model to restore missing regions using photographic logic and era-appropriate detail. - Steve’s Photo Conservator v2
A “do-no-harm” conservation approach with clearer damage taxonomy. Best when you want subtle repairs rather than dramatic restoration.
Writing, Summaries & Content Tools
These prompts help you work with text—extracting facts, cleaning up drafts, summarizing conversations, and turning research notes into readable prose.
Fact Extraction & Narrative
- Steve’s Fact Extractor v4
Extracts atomic factual claims from messy text into structured, citation-ready units. Built for genealogy notes and source digestion. - Steve’s Fact Narrator v4
Turns extracted facts into coherent narrative prose while preserving evidentiary boundaries and uncertainty markers.
Editing & Cleanup
- Quick Writing Cleanup v3
Lightweight cleanup for grammar, verbosity, and readability without altering meaning. - Steve’s Quick Editor v03
Fast editorial pass focused on clarity and flow, with a brief change summary. - Steve’s Content Decoder v01
Sense-making assistant for dense or ambiguous text. Clarifies meaning and returns structured highlights.
Summaries & Abstracts
- Summarize Chat v3
Summarizes a conversation into key points, themes, and actionable takeaways. Useful for turning research chats into archival abstracts. - Chat Conversation Abstractor v2
Produces a formal abstract: purpose, claims, evidence, and unresolved questions.
Specialized Tools
- Image Citation Builder v2
Generates consistent citations for images (especially archival photos) using standardized fields. - Linguistic Profiler v3
Profiles writing style—tone, lexical habits, rhythm, markers—to characterize voice or compare authorship across texts. - InfographicGPT v7
Converts complex source material into infographic-ready content with strong hierarchy and visual-friendly summarization.
Research Agents
These prompts configure AI to act as a research assistant with specific methodologies and quality standards.
- Research Agent Assignment v2.1 (recommended)
A reusable research-agent specification: subject classification, adaptive methodology, verification rules, structured deliverables, and quality gates. - Contract-First Genealogy System Prompt v3.1 (recommended)
System-level prompt for a GPS-compliant genealogy assistant. Enforces a contract-lock workflow, structured logs, and ethics-first research behavior. Use this when you want the AI to follow the Genealogical Proof Standard. - Research Assignment v0 (legacy)
Earlier deep-research framework. Available for reference.
GPT Configurations
Templates for building custom GPTs (for ChatGPT Plus/Team subscribers):
- Website Frontend GPT v3
Template for custom GPTs tied to a blog or site. Instructs domain-first search, summary generation, and menu-style follow-ups. - Open GeneaGPT Beta 0.4
Full identity and operating spec for Open GeneaGPT: role, tone, standards, citation posture, and response structure.
Audio Transcription Scripts
For those comfortable with Python, the repository includes scripts that wrap OpenAI’s Whisper API for transcribing audio recordings (oral histories, interviews, etc.):
- transcribe-4.py — Single-file transcription with selectable output formats
- transcribe-4o-chunk.py — Long-audio workflow that chunks, transcribes, and merges
- transcribe_v02.py — Minimal reference implementation
These require some technical setup. See the GitHub repository for details.
How to Use These Prompts
- Click a prompt link above to open it on GitHub
- Copy the full text (click the “Raw” button on GitHub for easy copying)
- Paste it into your AI tool (ChatGPT, Claude, Gemini, etc.)
- Provide your input — a document image, a photo to restore, text to process, or a research question
- Follow the output — most prompts produce structured results you can use directly
Tip: Many of these prompts work best when pasted at the start of a new conversation, setting the AI’s behavior for everything that follows.
License & Attribution
All prompts and tools are released under Creative Commons BY-NC-SA 4.0. You’re free to use, adapt, and share them for non-commercial purposes with attribution.
Full repository: github.com/DigitalArchivst/Open-Genealogy
Questions or suggestions? Contact me.