@sanity/ailf-studio 0.1.10 → 0.1.12

package/README.md

@@ -222,10 +222,11 @@ The plugin exports building blocks for custom views or extensions.
 
 ### Document Actions
 
-| Export                      | Description                                                                         |
-| --------------------------- | ----------------------------------------------------------------------------------- |
-| `GraduateToNativeAction`    | Converts a mirrored (read-only) task to a native (editable) task by removing origin |
-| `createRunEvaluationAction` | Factory for creating a Studio action that triggers evaluations                      |
+| Export                      | Description                                                                                  |
+| --------------------------- | -------------------------------------------------------------------------------------------- |
+| `GraduateToNativeAction`    | Converts a mirrored (read-only) task to a native (editable) task by removing origin          |
+| `RunTaskEvaluationAction`   | Triggers a pipeline evaluation scoped to a single task (registered on `ailf.task` documents) |
+| `createRunEvaluationAction` | Factory for creating a Studio action that triggers release-scoped evaluations                |
 
 ### Glossary
 

package/dist/index.d.ts

@@ -27,6 +27,45 @@ import { DocumentRef } from './document-ref.js';
 
 declare const GraduateToNativeAction: DocumentActionComponent;
 
+/**
+ * actions/RunTaskEvaluationAction.tsx
+ *
+ * Sanity Studio document action that triggers an AILF evaluation for a
+ * single task. Appears in the document action menu on `ailf.task` documents.
+ *
+ * ## Why this exists
+ *
+ * The release-level `RunEvaluationAction` answers "did my content changes
+ * help or hurt AI agents?". This task-level action answers "does this task
+ * definition actually work well?" — a tight authoring feedback loop for
+ * task authors iterating on prompts, rubrics, and canonical doc references.
+ *
+ * ## How it works
+ *
+ * Like `RunEvaluationAction`, this action does NOT hold any secrets. It
+ * creates an `ailf.evalRequest` document scoped to a single task. A
+ * server-side Sanity webhook picks up the document and dispatches the
+ * pipeline.
+ *
+ * ## Key differences from RunEvaluationAction
+ *
+ * - Scoped to a single task via `tasks: [taskId]`
+ * - No perspective/release — evaluates against the default doc source
+ * - Polls for reports matching the task's feature area
+ *
+ * @see packages/eval/src/webhook/eval-request-handler.ts
+ * @see .github/workflows/external-eval.yml
+ */
+
+/**
+ * Document action that triggers a single-task evaluation.
+ *
+ * Registered on `ailf.task` documents in the plugin definition.
+ * Creates an `ailf.evalRequest` doc with `tasks: [taskId]` to
+ * run all test cases for the task.
+ */
+declare const RunTaskEvaluationAction: DocumentActionComponent;
+
 declare function AssertionInput(props: ObjectInputProps): react_jsx_runtime.JSX.Element;
 
 declare function CanonicalDocInput(props: ObjectInputProps): react_jsx_runtime.JSX.Element;
@@ -309,6 +348,7 @@ declare const evalRequestSchema: {
     preview?: sanity.PreviewConfig<{
         perspective: string;
         status: string;
+        tasks: string;
     }, Record<string, unknown>> | undefined;
 };
 
@@ -731,11 +771,12 @@ interface TimelineDataPoint {
 
 /**
  * AILF Studio plugin — registers the report schema, dashboard tool,
- * and document actions (Graduate to Native for mirrored tasks).
+ * and document actions for `ailf.task` documents (Graduate to Native
+ * for mirrored tasks, Run Task Evaluation for single-task evals).
  *
  * This is the recommended way to install the plugin. It registers
  * schemas, the dashboard tool, and document-level actions in one call.
  */
 declare const ailfPlugin: sanity.Plugin<void>;
 
-export { AssertionInput, CanonicalDocInput, type ComparisonData, type ContentImpactItem, GLOSSARY, GraduateToNativeAction, MirrorBanner, type ProvenanceData, ReleasePicker, type ReportDetail, type ReportListItem, type RunEvaluationActionOptions, type ScoreItem, type SummaryData, SyncStatusBadge, type TimelineDataPoint, ailfPlugin, ailfTool, articleSearchQuery, comparisonPairQuery, contentImpactQuery, createRunEvaluationAction, distinctAreasQuery, distinctModesQuery, distinctPerspectivesQuery, distinctSourcesQuery, distinctTargetDocumentsQuery, evalRequestSchema, featureAreaSchema, latestReportsQuery, recentDocumentEvalsQuery, referenceSolutionSchema, reportDetailQuery, reportSchema, scoreTimelineQuery, taskSchema, webhookConfigSchema };
+export { AssertionInput, CanonicalDocInput, type ComparisonData, type ContentImpactItem, GLOSSARY, GraduateToNativeAction, MirrorBanner, type ProvenanceData, ReleasePicker, type ReportDetail, type ReportListItem, type RunEvaluationActionOptions, RunTaskEvaluationAction, type ScoreItem, type SummaryData, SyncStatusBadge, type TimelineDataPoint, ailfPlugin, ailfTool, articleSearchQuery, comparisonPairQuery, contentImpactQuery, createRunEvaluationAction, distinctAreasQuery, distinctModesQuery, distinctPerspectivesQuery, distinctSourcesQuery, distinctTargetDocumentsQuery, evalRequestSchema, featureAreaSchema, latestReportsQuery, recentDocumentEvalsQuery, referenceSolutionSchema, reportDetailQuery, reportSchema, scoreTimelineQuery, taskSchema, webhookConfigSchema };