Investigate a historical family question with a traceable research log

Proposed inputs

1. Name

   Question and known record copies

   Owner

   Researcher

   Minimum scope

   One historical event or relationship, authorized records and their locators.

   Quality check

   Separate transcriptions from images and mark illegible text.

2. Name

   Search boundary

   Owner

   Researcher

   Minimum scope

   Places, date range, permitted repositories and prior attempts.

   Quality check

   A website failure must remain different from a searched collection with no match.

Access and setup — proposed or unknown

Proposed minimum

1. Use a private working copy and keep every proposed tree change out of scope. Start with supplied records before requesting any new archive access.

Compatibility unknowns

1. Unknown: archive availability, indexing completeness, name variants and current account requirements.

External actions requiring approval

1. Approve repository searches, paid record orders, contact with archives, sensitive-data sharing and any later family-tree edits separately.

Why use this

1. An attractive family story can become hard to correct once an uncertain relationship enters a shared tree. The evidence trail needs to remain separable from the narrative.

2. Recording unsuccessful searches also helps a later researcher understand what was actually checked and what remained inaccessible.

Why an agent / simpler alternative

Useful judgment

Compare names, dates, places and informants while keeping incompatible identity candidates separate.

Simpler alternative

Inspect a known record manually when its identity and relevance are already settled.

Unsuitable when

The request involves living-person profiling, DNA interpretation, legal identity proof or restricted records.

Expected deliverable and review checks

Proposed artifact

A question-led ledger linking each claim to the exact record, search attempt and remaining identity doubt.

Critical checks

1. Two copies of the same original count as one evidence origin.

2. No match, inaccessible and not searched remain distinct states.

3. A person match must survive a place/date contradiction check; confidence percentages require a defensible method or remain absent.

Missing output evidence

No completed output or independently observed run was inspected in this research.

First check — proposed, unexecuted

Status

proposed_unexecuted

Synthetic fixture

Use fictional nineteenth-century records for two people named Mara Vale in different towns, plus a transcription copied from one original and one inaccessible register.

Acceptance checks

1. Retain both identity candidates and group the copied transcription with its origin.

2. Log the inaccessible register separately from a completed no-match search.

3. Return a targeted next-record question without changing a tree or declaring ancestry proved.

Stop conditions

1. Stop if living-person details, restricted access or a consequential identity decision enters scope.

Failure signals and proposed responses

1. Signal

   A common surname joins records from incompatible locations.

   Why it matters

   A plausible narrative may attach the wrong family.

   Proposed response

   Keep separate candidates and ask what record could distinguish them.

2. Signal

   No search result is treated as proof an event never happened.

   Why it matters

   Missing coverage can masquerade as negative evidence.

   Proposed response

   Record collection coverage and search limits; leave the event unresolved.

Effort and cost — unknown

Setup work

1. Define one question and organize locators for the records already held.

Ongoing review

1. The researcher verifies transcriptions and identity links before accepting a conclusion.

Unknowns

1. Unknown: archive charges, access restrictions and time needed to resolve ambiguities.

Limits

1. This review design does not certify genealogy standards, biological relationships or legal proof; unresolved evidence is an acceptable result.