ODATANO ASTRA

Analytics over Cardano and Midnight as OData V4. ASTRA reads the indexes the hosted ODATANO and NIGHTGATE instances already fill, aggregates them into minute, hour and day buckets and serves windows, series, rankings and anomalies under /odata/v4/astra. Every number names the block range it was computed from.

ASTRA runs hosted only, at api.odatano.dev, behind the same key as the other routes. A read costs one unit. There is no npm package; the raw-data services of ODATANO and NIGHTGATE stay reachable next to it.

Call it

# one row per chain: tip, lag, blocks and transactions 1 h, fees 24 h
curl https://api.preprod.odatano.dev/odata/v4/astra/ChainOverview \
  -H "Authorization: Bearer oda_..."

# every metric over one window, with the window before and the change in percent
curl "https://api.preprod.odatano.dev/odata/v4/astra/KeyFigures?\$filter=chain eq 'cardano' and window eq '7d'" \
  -H "Authorization: Bearer oda_..."

# fees per day, newest first: total, fee-paying transactions, average, median, p95, maximum
curl "https://api.preprod.odatano.dev/odata/v4/astra/FeesDaily?\$filter=chain eq 'midnight' and day ge 2026-09-01&\$orderby=day desc" \
  -H "Authorization: Bearer oda_..."

# the ten tokens with the most transfers today
curl "https://api.preprod.odatano.dev/odata/v4/astra/TopTokens?\$filter=chain eq 'cardano' and window eq '24h' and rank le 10" \
  -H "Authorization: Bearer oda_..."

# one metric over a rolling window, with the window before and the change in percent
curl "https://api.preprod.odatano.dev/odata/v4/astra/getWindow(chain='cardano',metric='tx.count',window='24h')" \
  -H "Authorization: Bearer oda_..."

# the same metric for Cardano and Midnight side by side
curl "https://api.preprod.odatano.dev/odata/v4/astra/compare(metric='fees.paid',window='7d')" \
  -H "Authorization: Bearer oda_..."

chain is cardano or midnight. Windows are 1h, 24h, 7d, 14d and 30d; rankings also take epoch. The entities take $filter (comparisons, and, or, not), $orderby, $top, $skip, $select and $count; a plain GET reads newest first. Every call is listed with an example response in the API reference under Analytics.

Ready-made entities

No metric id needed. Every entity holds both chains; a column a chain does not have is null, so Cardano and Midnight sit in one table.

Functions

Dimensions for rankings: token, policy, script, address, producer, pool, drep and tokenHolder. The last one needs a subject, the unit of the token whose holders are ranked. pool and drep read the newest epoch snapshot instead of a span of time.

Under the hood

The entities above are computed on read from ASTRA’s own bucket store: count, sum, min, max and a log histogram per metric, granularity and bucket start, plus per-entity counters behind the rankings and epoch snapshots for stake and governance. Those raw tables (MetricBuckets, EntityBuckets, Snapshots) and the aggregation cursor (SyncStatus) stay with the operator; the gateway does not offer them. ChainOverview carries what a consumer needs of the cursor: tip, both lags, lastBlockAt and backfilling.

Metrics

Metric ids are the same on Cardano and Midnight where the measure exists. kind decides the headline: a counter reports its sum, a distribution its average, a gauge its last value.

The full catalogue with units and descriptions is GET /odata/v4/astra/Metrics.

Two lags

ChainOverview reports two lags, and both matter when reading the recent windows. lagBlocks is how far ASTRA is behind its source, sourceLagBlocks how far the source is behind the chain. The windows are anchored to the wall clock, so while either lag is large the last hour looks empty. That is missing data, not a quiet chain. lastBlockAt says how far the numbers actually reach.

Reorg-safe and traceable

A reorg logged by the ODATANO or NIGHTGATE crawler marks the affected buckets invalid and the aggregator recomputes them. Every bucket carries fromHeight, toHeight and computedAt, so any number can be checked against the raw entities of /odata/v4/cardano-odata or /odata/v4/nightgate.

In SAP and Excel

The service is plain OData V4. Fiori elements, SAP Analytics Cloud, Excel and Power BI read BlocksDaily, FeesDaily, KeyFigures and ChainOverview like any other entity set, one row per day or per key figure, with named columns instead of metric ids:

GET /odata/v4/astra/TransactionsDaily?$filter=day ge 2026-09-01&$orderby=day desc
GET /odata/v4/astra/KeyFigures?$filter=window eq '30d'&$select=chain,metric,name,value,previous,changePct,unit

For AI agents

Both MCP servers carry these numbers as tools for their chain, registered when the gateway serves ASTRA: ODATANO-MCP for Cardano and NIGHTGATE-MCP for Midnight, same key, one unit a read.

"How busy was Cardano this week?"          → analytics_key_figures(window: "7d")
"Fees per day for the last two weeks"      → analytics_daily(topic: "fees", days: 14)
"Which tokens moved most today?"           → analytics_top(ranking: "tokens", window: "24h")
"How many Midnight transactions were shielded?" → analytics_key_figures(metric: "tx.shielded")
"Cardano and Midnight side by side"        → analytics_compare(metric: "tx.count", window: "7d")
"Anything unusual yesterday?"              → analytics_anomalies()

Next steps

• Get Started: Get a key and make the first calls
• MCP servers: The same key for AI agents, with the analytics_* tools on both chains
• API reference: Every call with an example response