| obligation_id | str | Unique id of this obligation. Carries the section marker and the printed page it begins on. |
| obligation_group_id | str | Id of the obligation group this obligation belongs to. The group is the unit a policy adopts. |
| content_id | str | Stable content address of the obligation's own words. Survives the regulator reissuing the document with different pagination; use it to join across releases. |
| framework | str | The regulatory framework this obligation belongs to. |
| regulator | str | The issuing authority. |
| jurisdiction | str | The jurisdiction the instrument applies in. |
| document_title | str | Title of the source instrument. |
| document_version | str | Version or issue date of the source instrument. |
| section_number | str | Section or provision marker as printed. |
| section_title | str | Heading of the section the obligation sits in. |
| subsection_title | str | Heading of the subsection, where the document prints one. |
| marker_path | str | The full marker path from the section to this obligation. |
| section_breadcrumb | str | Human-readable trail of headings above the obligation. |
| depth | int | Nesting depth of the obligation within the document structure. |
| pdf_page_start | int | PDF page the obligation begins on: the page number to type into a PDF reader. The PRINTED page appears inside obligation_id and both citation columns, and the two numberings can differ in instruments that are a slice of a larger book. |
| pdf_page_end | int | PDF page the obligation ends on. Differs from pdf_page_start on obligations that cross a page break. |
| verbatim_text | str | The regulator's words, byte-exact. THE authoritative field: cite this, never a derived field. |
| normalized_text | str | The same text with layout artifacts (line breaks, hyphenation) normalised for search. |
| obligation_paragraph | str | The full paragraph the obligation sits in. |
| context_window | str | The surrounding passage a reader needs to interpret the obligation. |
| preceding_paragraph | str | The paragraph printed before the obligation's own. |
| following_paragraph | str | The paragraph printed after the obligation's own. |
| governing_stem | str | The lead-in sentence that governs a list the obligation belongs to, where there is one. |
| modal | str | The normalised modal verb that makes this an obligation. |
| modal_strength | str | Mandatory, recommended or informational, read from the modal. |
| deontic_category | str | Obligation, prohibition, permission, recommendation, discouragement or informational. |
| is_negative | bool | True where the obligation is stated in the negative. |
| is_conditional | bool | True where the obligation applies only under stated conditions. |
| actor | str | Who the obligation falls on, in the document's own terms. Empty where the source names nobody. |
| action_required | str | What must be done, extracted from the obligation's own words. |
| object_of_obligation | str | What the action applies to. |
| applicability | str | Who or what the obligation covers, where the source states a scope. |
| frequency | str | How often, where the source states one. |
| timing | str | When or by when, where the source states one. |
| qualifications | str | Conditions and exceptions stated in the obligation's own words, as 'kind: text', pipe-separated. |
| functional_domains | str | GRC functional domains this obligation applies to, pipe-separated. |
| topics | str | Topic classification drawn from the instrument's own structure, pipe-separated. |
| definition_count | int | How many defined terms this obligation relies on. The definitions themselves, quoted from the document, are in the JSON delivery. |
| cross_reference_count | int | How many references this obligation makes to other provisions or instruments. The references themselves are in the JSON delivery. |
| legal_citation | str | Short citation for workpapers. |
| legal_citation_full | str | Full citation naming the instrument, section, printed page and regulator. |
| source_pdf_url | str | The official source. Free to retrieve; the authoritative publication. |