A traceable path from source to result.

01

Preserve the original record

Original downloaded bytes, retrieval timestamps, hashes, and parser versions remain separate from normalized records. A publication date is shown only when the source provides one. Retrieval time is not a legal effective date.

02

Keep facts separate from search forms

Names, scripts, aliases, identifiers, addresses, and uncertain dates retain their source meaning. Derived tokens and case folding support matching. They do not become facts asserted by the authority.

03

Retrieve candidates before scoring

Indexed candidate retrieval narrows the comparison set. Deterministic rules compare names and available identifiers, explain contributions, and report conflicts. Missing information does not prove a non-match.

04

Show the evidence behind a result

A potential match includes the source record, matched fields, understandable explanations, and the source and engine versions used. Similarity scores are not probabilities of wrongdoing.

05

Keep review decisions separate

Analyst decisions are recorded alongside the original outcome. A dismissal does not rewrite the source or erase the screening. Relevant source, subject, engine, or policy changes need a fresh review.

What this does not decide

A no-match result applies to the selected sources and dataset. It is not legal clearance. Ownership and control rules can apply to unlisted entities, and the relevant rules differ by jurisdiction. SanctionsKit does not claim validated ownership analysis, PEP screening, adverse-media screening, or transaction monitoring.

Vessel and aircraft attributes are preserved where provided by sources. Use the supported subject types shown in source discovery and compare identifiers against the original source record.