Once your sources have synced into the Enterprise Context Graph, the Records tab is where you browse the actual rows behind each node type, search for a specific one, and drill into a single record to see how it connects to everything around it. This page walks through browsing and search, what the status values mean, and what you'll find once you open a record — its relationship graph, its inspector, and its document preview.
Overview
Records shows one table per node type. Pick a node type from the picker — the first one is selected for you by default — then search and scroll; records load 20 at a time as you scroll down. Each column corresponds to a property, shown in the node's property order, and the primary-key column carries a PK pill in its header. Two system columns, Updated and Status, appear only when the records you've loaded actually carry those fields, so most node types won't show them at all.
Records here are browse-only — there's no sorting, filtering, editing, or deleting from this table, and rows are listed in a fixed internal order. Clicking a row opens that record's drill-down. If a node type has no data yet, you'll see "No records yet." Keep in mind the same empty message also appears if there's a loading problem, so if you're confident data has synced, refresh before assuming the node type is genuinely empty. Also note that a table column only shows a property if that property was marked queryable for the graph — a property that opted out shows as a dash even when the underlying source has a value for it.
Browsing and searching records
The search box searches across whichever of the node's properties have search enabled, matching on the server. It's a case-sensitive "contains" match, so "Acme" and "acme" won't return the same results, and list or object properties aren't included in search at all. Search text is capped at 256 characters. A search that finds nothing shows "No Results Found" with a Clear Search button to reset it, and while a new search is loading, the previous results stay visible for a moment instead of flashing to empty.
Two things to watch for: if none of a node's properties have search enabled, the search box will appear to do nothing, since every record still matches. And if your search results happen to lack the Status field, the Status column temporarily disappears until you clear the search. Switching to a different node type always clears whatever you'd searched for.
What the record statuses mean
Where a Status column is present, each record shows one of three values. Active means a golden record — the single merged, surviving version after quality checks, matching, and survivorship — was created, updated, or re-synced with no change. Flagged means the record failed a data quality check. Review covers everything else the pipeline produced, including records that were ultimately rejected — so Review doesn't necessarily mean a record is still in flight. To act on a problem record — see which rule flagged it, fix and resubmit a rejected record, or review a potential duplicate merge — you'll need the Ontology module's record screens for that same model rather than this view. One edge case: opening a record from a direct link before its page finishes loading shows no status or updated time yet.
List and object values
A property holding a list or an object renders as compact one-line text — simple lists are comma-joined, anything more structured shows as JSON. Hover the value to open a read-only, indented, copyable view of the full contents; clicking the value still opens the record rather than the hover view. Long plain-text values truncate with the full text available in a tooltip. This same treatment applies inside the record inspector as well.
Opening a record: the drill-down
Clicking a row replaces the table with a drill-down view: a header with a back arrow, the record's title and a type pill, then up to three panes — the relationship graph, an optional document preview (shown only when the record came from a file), and the inspector. The browser's Back button returns you to the record list, and the open record is reflected in the page URL, so you can share a link straight to it.
Exploring the relationship graph
Your record sits at the centre of the graph, surrounded by the records it connects to. A related record whose own data hasn't synced yet can show up as a bare placeholder carrying little more than its identifier — the relationship arrived before the record itself did. A legend maps each dot colour to a node type, with a "+N" chip if there are more types than fit in the legend. To keep dense graphs readable, only relationships touching the centre record, whatever's currently selected, and whatever you're hovering are drawn; edge colour shows direction relative to the record you opened, and a relationship's name appears when you hover over it. Hovering any record shows a summary card with its first few properties, the app it came from, and a count of its relationships. Selecting a record points both the inspector and the preview pane at it.
A record with hidden neighbours carries a +N pill — the record you opened starts with 25 already showing. The first click reveals 10 more, the next 25, then 50, doubling with each click after that. Revealed batches are spread across relationship types so no single type dominates a batch, and are ordered by connection strength, so the first reveal shows the record's strongest relationships rather than an arbitrary slice; reporting-line relationships are always shown in full up front. While a batch is loading, the pill shows "…". Once nothing is left hidden, the pill turns into a collapse control — except on the centre record, which can be expanded but never collapsed. Each expansion fetches one hop at a time, and an extremely well-connected record may not show every relationship in a single fetch.
The record inspector
The inspector sits alongside the graph and shows the selected record's title and type across two tabs. Props lists every property, primary key first, with PK and IDX pills shown on the centre record. Edges groups the record's outgoing and incoming relationships into separate lists; clicking one selects that record in the graph. The inspector currently shows only current values — where a value came from, or its history, isn't shown here yet.
Previewing source documents
Records derived from a file get a preview pane with Download and Maximize controls, plus a page selector for multi-page documents and zoom controls for images and PDFs. Images, video, PDF, CSV, HTML, JSON, XML, plain text, and VTT subtitle or transcript files render inline, with text-based formats opening read-only and offering a copy action. Anything else — Word or Excel files, archives, and similar formats — shows "Preview not available," with a prompt to download the file to view its contents. If the file itself can't be fetched, you'll see "Couldn't load this document" instead. Records with no attached file simply have no preview pane at all.
Notes
Node types without a primary key can't offer stable deep links to an individual record — the link identifies a record by its position in the list, which can shift as data changes.
Records in this view are read-only. To fix, resubmit, or review flagged and duplicate records, use the Ontology module's record screens for the same model.
A "No records yet" message can also mean a loading problem rather than truly empty data — refresh before concluding there's nothing there.