astrupdata

Changelog

Release history and platform updates.

0.3.02026-07-26

Settlement Routes

  • ·New section under Analytics: which parties a transfer was observed to settle through, derived from the counterparties of each settlement. Keeta settles atomically — measured on mainnet, 53 of 53 multi-operation transactions had every leg in the same block — so a route cannot be read from block ordering. It comes from who the operations moved value between, which was already in the indexed data.
  • ·Four pages. Overview charts throughput, value, live routes and route-class activity, with a sortable table of every observed path. Parties is a master-detail view per account: an origin-to-destination matrix, per-account small multiples on a shared scale, and inbound / outbound / round-trip over time. Routing Graph shows what the anchor resolver declares the network can route. Method carries every caveat, once.
  • ·Backfilled to 90 days. The route aggregation held 31 days; the underlying transactions held 149, so the missing span was recomputed with the same function the hourly job uses.
  • ·Every declared route that can touch Keeta is listed — 226 of the registry's 4,105 edges, grouped into 70 provider corridors. The other 3,879 belong to a provider that declares no Keeta location and can never appear in settlement data, so listing them as potential routes here would not have been true.

Workbench

  • ·A new tool at Charts Library → Workbench: pick a measure, a breakdown, a resolution and a transform, and the chart appears. Two sources — settlement routes and network volume — with the whole configuration in the URL, so a chart you build is a link you can send.
  • ·Transforms run over data already fetched: cumulative, moving average, rebase to 100 and change per period. Rebasing indexes each series against its own first reading, so series of very different sizes can be compared by movement.
  • ·It offers only combinations the stored data can answer. Per-token breakdown, per-provider breakdown, hop counts and distribution statistics are absent because the aggregation does not hold them, and the reason is recorded next to each.

Corrections

  • ·A community token declared itself "USDT" and was presented as Tether. It was swapping 18 million of itself against $16 of KTA. Two paths allowed it: the indexer took a token's own on-chain metadata at face value, and the symbol resolver mapped anything named "usdt" or "tether" to USDT. Neither checked an address. USDT and PYUSD have been removed from the codebase entirely — they were carried in eleven places with no address anywhere, which is a placeholder any token can occupy. A symbol is now confirmed only by a registered address, and a token claiming a confirmed symbol without being that address is marked rather than believed.
  • ·Bridge withdrawals were invisible in route data. The escrow vault was filtered out as bookkeeping, but a withdrawal sends user to vault, so removing the vault left nothing to route. Measured: 68 vault sends in 400 consecutive transactions, none of them counted.
  • ·The route mix chart rendered in the bottom 1% of its own frame. It computed percentages and then asked the chart to normalise them again, against an axis pinned to 0-100. Replaced with unstacked counts, which also show a class going quiet — a share chart cannot.
  • ·The headline figure above every multi-series chart ignored the series you selected. It was computed from the first key of everything fetched, so on Fiat Composition, picking CAD showed HKD's value. It now follows the selection, and shows nothing when several series are drawn rather than silently speaking for one.
  • ·Sent and Received misreported every swap desk. A round trip leaves and arrives at the same party, so it was counted on both sides — the FX resolver read 449 out and 419 in when it is really 86 out, 56 in and 361 round trips.
  • ·Route classification keyed on display labels and worked by coincidence: one account's label happened to contain the word "bridge". Renaming that account would have silently reclassified its traffic. It now keys on a frozen identifier.
  • ·Coverage was removed rather than annotated. It read as a completeness score and could never reach 100%: of 1,997 unrouted transactions in a 3,000 sample, 1,324 are plain wallet-to-wallet sends that pass through no infrastructure and so have no route to show.

Interface

  • ·The sidebar uses one type system. Group headings were small tracked capitals sitting directly above sentence-case items; they now match the items, with weight and colour carrying the hierarchy.
  • ·Chart toolbar labels dropped their tracked capitals for the same reason — one word in capitals among sentence-case siblings reads as a different system rather than as emphasis.
  • ·Section cards on the landing page and Terminal Home lost the small grey tag line under each description. It restated the sentence above it in fewer words.
0.2.02026-07-25

Terminal

  • ·Reorganised the product around a new information architecture: a marketing landing page at the root, and the whole application under /terminal with permanent redirects from every previous URL.
  • ·New sidebar as a catalog tree — Home, Dashboards, Charts Library and Explorer — with collapsible groups that remember their state.
  • ·Light theme is now the default for first-time visitors.

Charts Library

  • ·Added a searchable catalog with a dedicated page per metric, driven by a central metric registry and rendered on a new ECharts engine with pan, zoom, crosshair, log scale and PNG export.
  • ·Rebuilt Fiat On-Chain as a master-detail view with per-currency series in native units.
  • ·Line charts are now filled, matching the area charts; multi-series charts stay unfilled so crossings remain readable.

Data integrity

  • ·Fixed nine paths where the transaction indexer could stop reading early and still advance its cursor. Anything skipped fell behind the cursor and was never re-read — silently, while the run reported success. The scan must now prove it covered the window before the cursor moves, and fails visibly when it cannot.
  • ·Timestamps are no longer invented. A staple whose timestamp could not be read was previously stamped with the current time, which could push the cursor into the future and stop indexing permanently while every health metric still read green. Such rows are now recorded separately and never touch the cursor.
  • ·The reconciliation watchdog no longer reports a clean window when it read an incomplete one. It shared the same defects as the code it audits, so a decode failure was invisible on both sides of its comparison. Its first supervised run confirmed 324 staples on chain against 324 indexed.
  • ·Transactions that arrive after their hour has closed are now counted. They were previously dropped by both the live and the batch aggregation layer.
  • ·A backlog can now drain. The scan keeps what it read and resumes where it stopped, instead of restarting from the chain tip each minute and never finishing.
  • ·Repaired daily active addresses for 8–10 July, where a failed read had left 9 July recorded as zero despite 631 real transactions that day.

Fixes

  • ·Daily Swap Volume, Cumulative Swap Volume and Community Token Prices rendered blank; their stored time labels carried no year and could not be read back as dates.
  • ·The Daily Active Addresses chart no longer dips below its axis, and the Wallet Tiers change column now follows the selected range instead of always showing 30 days.
  • ·The Base chip is now identical on the landing page and in the live feed.
  • ·Panels that cannot read their data now say so, instead of showing an empty state that looks like a healthy zero.
0.1.122026-05-28

Summary

  • ·Added a fiat dimension for Keeta Personal with three new analytics pages: Fiat On-Chain (per-currency supply in native units), Fiat Flow (deposit/withdrawal ramp), and Fiat Composition (book registry + share-over-time).
  • ·Corrected on-chain parsing so fiat deposit/withdrawal flow is attributed and counted correctly; previously these operations were silently dropped, leaving ramp flow blank.
  • ·Added a Fiat Volume KPI and a dedicated Fiat Volume chart on Transaction Volume; reframed Volume by Asset Class as share-of-period to avoid misleading scale.
  • ·Pre-aggregated the fiat supply series so the Fiat On-Chain page reads a single document per visit.
0.1.112026-05-01

Summary

  • ·Hardened transaction indexing so cursors advance only after dependent data writes complete.
  • ·Corrected aggregate data handling for active addresses, community token totals, and transaction filtering.
0.1.102026-04-28

Summary

  • ·Improved bridge reconciliation, explorer/indexer visibility, and analytics presentation.
  • ·Reduced noisy NVT and velocity interpretation so the metrics read more neutrally.
0.1.92026-04-28

Summary

  • ·Aligned bridge classification across explorer, dashboard, and scheduled data jobs.
  • ·Added rebuild and regression coverage for bridge and resolver data flows.
0.1.82026-04-24

Summary

  • ·Improved Explorer loading states and empty-state messaging.
  • ·Hardened transaction indexing, pagination, dedupe, and pipeline reliability.
  • ·Moved more dashboard data to indexed and pre-aggregated sources.
0.1.72026-04-19

Summary

  • ·Rolled out the lighter UI system across key product pages.
  • ·Added FX Liquidity and improved market/dashboard readability.
  • ·Improved mobile layout behavior across explorer and swaps.
0.1.62026-04-12

Summary

  • ·Updated certification, NVT, and velocity pipelines.
  • ·Cleaned up deployment and navigation behavior.
0.1.52026-04-10

Summary

  • ·Mobile-first UI pass across dashboard, analytics, explorer, and market views.
  • ·General layout, readability, and consistency improvements.
0.1.42026-03-25

Summary

  • ·Expanded swap volume tracking and Network Status metrics.
  • ·Improved dashboard summaries, bridge flow views, and supporting data infrastructure.
0.1.32026-03-20

Summary

  • ·Moved pricing and bridge widgets toward Firestore-backed data.
  • ·Improved audit visibility and corrected several metric-labeling issues.
0.1.22026-03-19

Summary

  • ·Improved bridge, TOCV, capital flow, and token market data accuracy.
  • ·Expanded internal reporting and screenshot views.
0.1.12026-03-16

Summary

  • ·Added admin screenshot views and X profile links.
0.1.02026-03-15

Summary

  • ·Added liquidity token indicators and token market cap improvements.
@astrupdata