@kortyx/telemetry-contracts 0.17.2 → 0.19.0

package/CHANGELOG.md

@@ -1,5 +1,36 @@
 # Changelog
 
+## [0.19.0](https://github.com/kortyx-io/kortyx/compare/telemetry-contracts-v0.18.0...telemetry-contracts-v0.19.0) (2026-10-18)
+
+
+### Features
+
+* **studio:** add sample project and guided onboarding ([#363](https://github.com/kortyx-io/kortyx/issues/363)) ([25da0b9](https://github.com/kortyx-io/kortyx/commit/25da0b9c92072446a7503b23ba4f157aebec8185))
+* **studio:** persist table preferences with scoped browser caches ([#367](https://github.com/kortyx-io/kortyx/issues/367)) ([9b84295](https://github.com/kortyx-io/kortyx/commit/9b8429563d0a065939381de5afe1c62e762157d0))
+
+
+### Dependencies
+
+* The following workspace dependencies were updated
+  * dependencies
+    * @kortyx/prompts bumped to 0.2.3
+
+## [0.18.0](https://github.com/kortyx-io/kortyx/compare/telemetry-contracts-v0.17.2...telemetry-contracts-v0.18.0) (2026-10-16)
+
+
+### Features
+
+* **evals:** configurable deadlines and original error visibility ([#354](https://github.com/kortyx-io/kortyx/issues/354)) ([7dde173](https://github.com/kortyx-io/kortyx/commit/7dde173b54eb37a2d28ff28ac342a3333e5ed902))
+* **studio:** standardize tables, filters and column behavior ([#353](https://github.com/kortyx-io/kortyx/issues/353)) ([c662e59](https://github.com/kortyx-io/kortyx/commit/c662e59d814a6a9f12c709595968ee7b1ece3191))
+
+
+### Dependencies
+
+* The following workspace dependencies were updated
+  * dependencies
+    * @kortyx/prompts bumped to 0.2.2
+    * @kortyx/core bumped to 0.15.0
+
 ## [0.17.2](https://github.com/kortyx-io/kortyx/compare/telemetry-contracts-v0.17.1...telemetry-contracts-v0.17.2) (2026-10-13)
 
 

package/README.md

@@ -1,3 +1,43 @@
 # @kortyx/telemetry-contracts
 
 Shared Zod wire contracts and fixtures for Kortyx telemetry producers and consumers.
+
+Studio tables share `TableFiltersSchema`, `TableFilterRuleSchema`, the field
+registries and validation/matching helpers. Groups select `match: "all" | "any"`;
+values within a rule are alternatives. Text comparison is case-insensitive and
+literal (wildcards are not query syntax). Negative comparisons require a value;
+`missing` explicitly selects absent data. Pass field definitions to the matcher
+so date equality compares instants, while text equality compares literal text.
+
+Limits: 32 rules, 30 values per rule, 512 characters per string, 32 KiB serialized
+filters. Dates require ISO timestamps with an offset. Unknown fields, unsupported
+operators, malformed values and reversed ranges fail validation. Legacy Studio
+query aliases remain readable through `readStudioTableFilters`.
+
+Column registries include execution identity, environment, completion timestamps,
+prompt assignments and evaluation judge metadata. `tablePassRate` returns a
+percentage of recorded passed/failed/error cases (unrun cases are excluded), or
+no value when none finished. `tableQueueMs` returns elapsed queue milliseconds
+only when a valid start timestamp is recorded.
+
+Studio preference contracts include `StudioTablePreferencesSchema` (database
+snapshot with revision), `StudioTablePreferenceWriteSchema` (scope/revision plus
+one defaults/upsert-view/delete-view operation), and
+`StudioTableConfigurationSchema` (portable schema-version-1 configuration without
+owner IDs or database revision). Stable table IDs are registered centrally.
+Layouts allow at most 256 column keys, finite widths of 1–10000 px, and left/right
+pins. There are at most 30 uniquely identified views per table. View names are
+1–80 characters and filter/scope strings remain bounded. APIs derive ownership
+from authenticated principals; serialized configuration cannot grant access.
+
+### Studio sample identities
+
+`SAMPLE_PROJECT` exports the versioned sample project's stable record IDs,
+recording/reference timestamps, and future guided-tour anchors. Sample records
+are fictional and read-only; these identifiers are not authentication scopes.
+Dataset content is maintained by Studio and the internal telemetry DB package.
+
+The sample manifest also includes stable IDs for the two-turn session and the
+saved evaluation baseline. Dataset version changes preserve existing identities;
+Studio chapters resolve supported views from those IDs rather than persisting
+arbitrary resource paths.