MCP server for GEDCOM genealogical data manipulation, validation, and analysis
GedFire demonstrates solid tool design with clear naming (verb_noun pattern), comprehensive descriptions (avg 180 chars), and well-structured JSON schemas. All 7 tools have input schemas with typed parameters and descriptions. Tool names are action-oriented (validate_, apply_, find_, get_, date_calc). Descriptions explain WHAT the tool does and WHEN to use it. However, output schemas are not explicitly documented in the source code provided, and error handling guidance is minimal. The validate_changeset/apply_changeset pair shows good composition (validate before destructive apply). Parameter descriptions are generally adequate but lack explicit format constraints (e.g., date format examples, xref pattern). No evidence of enum constraints for operation types in date_calc.
Apply a proposal changeset to this server's bound GEDCOM document: validate every selected op, apply it to an in-memory copy, verify the result (byte-stable round-trip, pointer resolution, record-count deltas, no newly created source left uncited), and only then write the file -- there is no window where a bad state exists on disk. Call validate_changeset first with the same arguments to preview without writing. Refuses to run at all when this server was started with --read-only.
Perform genealogical date calculations: normalize dates to canonical form, add or subtract age intervals, or compute the difference between two dates.
Search for a person in the bound GEDCOM document by name, returning either a single matched person with full details (families, events, sources), a list of candidate matches with summary information, or suggestions for names that did not match.
Get summary statistics about the bound GEDCOM document: person count, family count, GEDCOM version, and GedFire version.
Retrieve a complete record from the bound GEDCOM document by cross-reference (xref): person records with all events, sources, media, notes, families, and restrictions; family records with spouses, children, and marriage details.
Output schemas not documented in source. Tools return structured results (CallToolResults.Success with JsonElement) but no schema definitions visible for what fields agents should expect.
date_calc operation parameter lacks enum constraint. Description says 'One of: normalize, add, sub, diff' but no formal enum in schema, LLMs may hallucinate invalid operations.
Parameter format constraints missing. date_calc 'date' and 'age' parameters lack explicit GEDCOM format examples or regex patterns in descriptions. xref parameter in get_record lacks pattern documentation.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 53 | 2025-06-18+ | v2 |
Validate a proposal changeset against this server's bound GEDCOM document: parse the changeset, validate every selected op against the current document state, apply it to an in-memory copy, and verify the result (byte-stable round-trip, pointer resolution, record-count deltas, no newly created source left uncited) — all without writing the file. Call this first to preview before using apply_changeset.
Validate the bound GEDCOM document against the GEDCOM specification: check for structural errors, warnings, and constraint violations. Returns a pass/fail result with detailed diagnostics.
Error handling guidance minimal. Tests show error responses (e.g., 'items must be...') but no recovery hints. When validate_changeset fails, agent doesn't know whether to retry, adjust parameters, or escalate.
find_person nicknames parameter is optional JSON object but no schema structure documented. LLMs cannot infer the expected shape (e.g., {"nickname": "canonical"}).