An observatory
for your data.
Elara is a local-first PostgreSQL workbench. Precise, calm, and entirely yours — no account, no cloud, no noise.
Free · macOS & Windows
Built to be looked through,
not at.
A schema-aware SQL editor with real completion. A result grid that streams large sets without flinching. Staged edits that show you their SQL before anything touches the database.
1WITH carrier_performance AS (
2 SELECT
3 s.carrier,
4 COUNT(s.id) AS total_shipments,
5 COUNT(s.id) FILTER (WHERE s.status = 'delivered') AS delivered_count,
6 COUNT(s.id) FILTER (WHERE s.status = 'in_transit') AS in_transit_count,
7 COUNT(s.id) FILTER (WHERE s.status = 'label_created') AS pending_count,
8 ROUND(AVG(EXTRACT(EPOCH FROM (s.shipped_at - s.created_at)) / 3600)::numeric, 2) AS avg_fulfillment_hours
9 FROM commerce.shipments s
10 WHERE s.created_at >= '2026-08-01 00:00:00+00'::timestamptz
11 GROUP BY s.carrier
12)
13SELECT
14 cp.carrier,
15 cp.total_shipments,
16 cp.delivered_count,
17 cp.in_transit_count,
18 cp.pending_count,
19 ROUND((cp.delivered_count::numeric / NULLIF(cp.total_shipments, 0)) * 100, 1) || '%' AS delivery_rate,
20 COALESCE(cp.avg_fulfillment_hours, 0.00) AS avg_dispatch_hrs,
21 DENSE_RANK() OVER (ORDER BY cp.delivered_count DESC, cp.avg_fulfillment_hours ASC) AS rank
22FROM carrier_performance cp
23ORDER BY rank ASC;
| carriertext | total_shipmentsbigint | delivered_countbigint | in_transit_countbigint | pending_countbigint | delivery_ratetext | avg_dispatch_hrsnumeric | rankbigint | |
|---|---|---|---|---|---|---|---|---|
| 1 | FedEx | 3,842 | 3,680 | 112 | 50 | 95.8% | 4.12 | 1 |
| 2 | DHL | 3,120 | 2,985 | 90 | 45 | 95.7% | 4.85 | 2 |
| 3 | UPS | 2,750 | 2,610 | 85 | 55 | 94.9% | 5.10 | 3 |
| 4 | USPS | 1,680 | 1,540 | 70 | 70 | 91.7% | 6.40 | 4 |
-- Socket reset during statement cancellation
-- PostgreSQL transaction status unconfirmed
delete from audit_logs
where created_at < now() - '90 days'::interval;
Honest when things
go wrong.
When a network socket resets or a cancellation is in flight, most database tools assume success and show a green checkmark. Elara refuses to guess.
If connection loss makes the outcome of a commit, rollback, or cancellation unknowable, Elara explicitly marks the transaction as uncertain. It preserves your query, explains what remains true, and protects your database from silent mutations until PostgreSQL confirms the server state.
What Elara won't do.
- Chase driver counts at the expense of workflow depth.
- Require the cloud for local database work.
- Route your database traffic through our servers.
- Gate the core editor behind paid AI.