Inspector, Dashboard & health

The Inspector dashboard is the first thing you land on. Its job is to let an operator answer four questions at a glance: is the instance healthy, how much state is in it, what's been happening recently, and who is writing to it.

Each panel is read-only and reflects live data from the Neotoma API, the same numbers GET /stats reports for programmatic consumers.

Stat row

The top row counts the four primitive Neotoma rows: entities, observations, relationships, and sources. Numbers come straight from GET /stats and reflect the operator's user_id scope. Trend hints ("+128 today") are derived from the same endpoint's daily aggregates and update without polling spam.

Entities by type

The histogram is the cardinality of each registered entity_type, not its schema field width. This is the same number the REST API returns under entities_by_type on /stats, and is the canonical source for "how many of each type" questions (don't substitute list_entity_types.field_count, which measures schema width).

Attribution coverage

The attribution panel surfaces what fraction of recent writes were signed at each AAuth tier, hardware, software, unverified_client, and anonymous. A healthy production instance trends toward hardware/software; a high anonymous share usually means clients are sending generic clientInfo.name values that get normalised away. See AAuth for the tier definitions.

Recent activity

Below the panels, a recent timeline strip lists the latest store / correct / link / merge events with their entity, agent, and trust tier. Click-through opens the corresponding entity detail; right-click copies the row's entity_id.

System health

A health card pings /healthz and /storage/health and reports SQLite integrity, write-ahead log depth, and last successful backup. SQLite corruption surfaces here first, followed by the operator-targeted recovery prompt described in Troubleshooting.