@sanity/ailf 0.4.1 → 1.0.0

package/dist/_vendor/ailf-core/config-helpers.d.ts

@@ -0,0 +1,174 @@
+/**
+ * define* identity helpers — typed configuration authoring functions.
+ *
+ * Each function is a **pure identity function** that returns its input
+ * unchanged. The value is purely at the type level: TypeScript narrows
+ * the parameter type so authors get full IDE autocomplete and compile-
+ * time validation when writing configuration files.
+ *
+ * Runtime validation is NOT performed here — it happens later in the
+ * pipeline via Zod schemas. This separation keeps the authoring
+ * experience fast (no schema import overhead at edit time) and lets
+ * the pipeline control when and how validation errors are surfaced.
+ *
+ * Usage:
+ * ```typescript
+ * import { defineTask, env } from "@sanity/ailf"
+ *
+ * export default defineTask({
+ *   id: "groq-projection-basics",
+ *   mode: "literacy",
+ *   title: "GROQ Projection Basics",
+ *   // ... full autocomplete here
+ * })
+ * ```
+ *
+ * @see docs/design-docs/architecture-overhaul/typescript-configuration.md (canonical)
+ */
+import type { EvalConfig } from "./schemas/eval-config.js";
+import type { FeatureRegistry, RubricConfig, ThresholdConfig } from "./schemas/pipeline.js";
+import type { SchedulesFile } from "./schemas/schedules.js";
+import type { SinksFile } from "./schemas/sinks.js";
+import type { ModelsConfig } from "./types/index.js";
+import type { GeneralizedTaskDefinition } from "./types/generalized-task.js";
+/**
+ * Define an AILF evaluation configuration.
+ *
+ * Used in `ailf.config.ts` files for typed configuration authoring.
+ */
+export declare function defineConfig(config: EvalConfig): EvalConfig;
+/**
+ * Define an evaluation task with full type narrowing by mode.
+ *
+ * Setting `mode: "literacy"` narrows available fields to literacy-specific
+ * options; `mode: "mcp-server"` narrows to MCP-specific options; etc.
+ *
+ * The returned value is the input unchanged — this is a type-only helper.
+ */
+export declare function defineTask(task: GeneralizedTaskDefinition): GeneralizedTaskDefinition;
+/**
+ * Define the model registry (models to evaluate and grader model).
+ *
+ * Used in `config/models.ts` for typed model configuration.
+ */
+export declare function defineModels(models: ModelsConfig): ModelsConfig;
+/**
+ * Define rubric templates and scoring profiles.
+ *
+ * Used in `config/rubrics.ts` for typed rubric configuration.
+ */
+export declare function defineRubrics(rubrics: RubricConfig): RubricConfig;
+/**
+ * Define quality thresholds for readiness gates and regression alerts.
+ *
+ * Used in `config/thresholds.ts` for typed threshold configuration.
+ */
+export declare function defineThresholds(thresholds: ThresholdConfig): ThresholdConfig;
+/**
+ * Define the product feature coverage registry.
+ *
+ * Used in `config/features.ts` for typed feature tracking.
+ */
+export declare function defineFeatures(features: FeatureRegistry): FeatureRegistry;
+/**
+ * Define report delivery sinks (Slack, BigQuery, webhooks).
+ *
+ * Used in `config/sinks.ts` for typed sink configuration.
+ */
+export declare function defineSinks(sinks: SinksFile): SinksFile;
+/**
+ * Define scheduled evaluation cron jobs and digest config.
+ *
+ * Used in `config/schedules.ts` for typed schedule configuration.
+ */
+export declare function defineSchedules(schedules: SchedulesFile): SchedulesFile;
+/**
+ * Source configuration — typed inline until a dedicated schema exists.
+ *
+ * Sources define where documentation is fetched from (production, branch,
+ * local). The shape matches config/sources.yaml entries.
+ */
+export interface SourceEntry {
+    /** Source name (e.g., "production", "branch", "local") */
+    name: string;
+    /** Base URL for the documentation site */
+    baseUrl: string;
+    /** Sanity project ID */
+    projectId?: string;
+    /** Sanity dataset */
+    dataset?: string;
+    /** Content release perspective */
+    perspective?: string;
+    /** Whether this is the default source */
+    default?: boolean;
+}
+/**
+ * Define documentation source configurations.
+ *
+ * Used in `config/sources.ts` for typed source configuration.
+ */
+export declare function defineSources(sources: SourceEntry[]): SourceEntry[];
+/**
+ * Prompt template configuration — typed inline until a dedicated schema exists.
+ */
+export interface PromptEntry {
+    /** Unique identifier (used in Promptfoo config references) */
+    id?: string;
+    /** Human-readable label (shown in results display) */
+    label?: string;
+    /** Prompt template name */
+    name: string;
+    /** Template string with `{{variable}}` placeholders */
+    template: string;
+    /** System message (for chat-format prompts) */
+    systemMessage?: string;
+    /** Supported modes for this prompt */
+    modes?: string[];
+}
+/**
+ * Define prompt templates.
+ *
+ * Used in `config/prompts.ts` for typed prompt configuration.
+ */
+export declare function definePrompts(prompts: PromptEntry[]): PromptEntry[];
+/**
+ * A preset is a named bundle of configuration that can be referenced
+ * by tasks or suites to apply shared settings.
+ */
+export interface PresetConfig {
+    /** Preset name */
+    name: string;
+    /** Default providers for tasks using this preset */
+    providers?: string[];
+    /** Default rubric template */
+    rubric?: string;
+    /** Default assertion templates */
+    assertions?: unknown[];
+    /** Default options */
+    options?: Record<string, unknown>;
+}
+/**
+ * Define reusable configuration presets.
+ *
+ * Used in `config/presets.ts` for typed preset configuration.
+ */
+export declare function definePreset(preset: PresetConfig): PresetConfig;
+/**
+ * A pricing table entry for cost estimation.
+ */
+export interface PricingEntry {
+    /** Model ID (e.g., "openai:gpt-4o") */
+    modelId: string;
+    /** Cost per 1K input tokens in USD */
+    inputPer1k: number;
+    /** Cost per 1K output tokens in USD */
+    outputPer1k: number;
+    /** Optional label */
+    label?: string;
+}
+/**
+ * Define a pricing table for cost estimation.
+ *
+ * Used in `config/pricing.ts` for typed pricing configuration.
+ */
+export declare function definePricingTable(entries: PricingEntry[]): PricingEntry[];

package/dist/_vendor/ailf-core/config-helpers.js

@@ -0,0 +1,150 @@
+/**
+ * define* identity helpers — typed configuration authoring functions.
+ *
+ * Each function is a **pure identity function** that returns its input
+ * unchanged. The value is purely at the type level: TypeScript narrows
+ * the parameter type so authors get full IDE autocomplete and compile-
+ * time validation when writing configuration files.
+ *
+ * Runtime validation is NOT performed here — it happens later in the
+ * pipeline via Zod schemas. This separation keeps the authoring
+ * experience fast (no schema import overhead at edit time) and lets
+ * the pipeline control when and how validation errors are surfaced.
+ *
+ * Usage:
+ * ```typescript
+ * import { defineTask, env } from "@sanity/ailf"
+ *
+ * export default defineTask({
+ *   id: "groq-projection-basics",
+ *   mode: "literacy",
+ *   title: "GROQ Projection Basics",
+ *   // ... full autocomplete here
+ * })
+ * ```
+ *
+ * @see docs/design-docs/architecture-overhaul/typescript-configuration.md (canonical)
+ */
+// ---------------------------------------------------------------------------
+// Config-level helpers
+// ---------------------------------------------------------------------------
+/**
+ * Define an AILF evaluation configuration.
+ *
+ * Used in `ailf.config.ts` files for typed configuration authoring.
+ */
+export function defineConfig(config) {
+    return config;
+}
+// ---------------------------------------------------------------------------
+// Task-level helpers
+// ---------------------------------------------------------------------------
+/**
+ * Define an evaluation task with full type narrowing by mode.
+ *
+ * Setting `mode: "literacy"` narrows available fields to literacy-specific
+ * options; `mode: "mcp-server"` narrows to MCP-specific options; etc.
+ *
+ * The returned value is the input unchanged — this is a type-only helper.
+ */
+export function defineTask(task) {
+    return task;
+}
+// ---------------------------------------------------------------------------
+// Model registry helpers
+// ---------------------------------------------------------------------------
+/**
+ * Define the model registry (models to evaluate and grader model).
+ *
+ * Used in `config/models.ts` for typed model configuration.
+ */
+export function defineModels(models) {
+    return models;
+}
+// ---------------------------------------------------------------------------
+// Rubric helpers
+// ---------------------------------------------------------------------------
+/**
+ * Define rubric templates and scoring profiles.
+ *
+ * Used in `config/rubrics.ts` for typed rubric configuration.
+ */
+export function defineRubrics(rubrics) {
+    return rubrics;
+}
+// ---------------------------------------------------------------------------
+// Threshold helpers
+// ---------------------------------------------------------------------------
+/**
+ * Define quality thresholds for readiness gates and regression alerts.
+ *
+ * Used in `config/thresholds.ts` for typed threshold configuration.
+ */
+export function defineThresholds(thresholds) {
+    return thresholds;
+}
+// ---------------------------------------------------------------------------
+// Feature registry helpers
+// ---------------------------------------------------------------------------
+/**
+ * Define the product feature coverage registry.
+ *
+ * Used in `config/features.ts` for typed feature tracking.
+ */
+export function defineFeatures(features) {
+    return features;
+}
+// ---------------------------------------------------------------------------
+// Sink helpers
+// ---------------------------------------------------------------------------
+/**
+ * Define report delivery sinks (Slack, BigQuery, webhooks).
+ *
+ * Used in `config/sinks.ts` for typed sink configuration.
+ */
+export function defineSinks(sinks) {
+    return sinks;
+}
+// ---------------------------------------------------------------------------
+// Schedule helpers
+// ---------------------------------------------------------------------------
+/**
+ * Define scheduled evaluation cron jobs and digest config.
+ *
+ * Used in `config/schedules.ts` for typed schedule configuration.
+ */
+export function defineSchedules(schedules) {
+    return schedules;
+}
+/**
+ * Define documentation source configurations.
+ *
+ * Used in `config/sources.ts` for typed source configuration.
+ */
+export function defineSources(sources) {
+    return sources;
+}
+/**
+ * Define prompt templates.
+ *
+ * Used in `config/prompts.ts` for typed prompt configuration.
+ */
+export function definePrompts(prompts) {
+    return prompts;
+}
+/**
+ * Define reusable configuration presets.
+ *
+ * Used in `config/presets.ts` for typed preset configuration.
+ */
+export function definePreset(preset) {
+    return preset;
+}
+/**
+ * Define a pricing table for cost estimation.
+ *
+ * Used in `config/pricing.ts` for typed pricing configuration.
+ */
+export function definePricingTable(entries) {
+    return entries;
+}

package/dist/_vendor/ailf-core/env-helper.d.ts

@@ -0,0 +1,35 @@
+/**
+ * env() — Typed environment variable access for configuration files.
+ *
+ * Provides a statically-analyzable alternative to raw `process.env` access.
+ * Tooling can detect which environment variables a config file depends on
+ * by scanning for `env("VAR_NAME")` calls.
+ *
+ * Usage:
+ * ```typescript
+ * import { env, defineConfig } from "@sanity/ailf"
+ *
+ * export default defineConfig({
+ *   source: {
+ *     projectId: env("SANITY_PROJECT_ID", "3do82whm"),
+ *     dataset: env("SANITY_DATASET", "next"),
+ *     apiToken: env("SANITY_API_TOKEN"), // throws if missing
+ *   },
+ * })
+ * ```
+ *
+ * YAML configs continue using the existing `${{ VAR | default }}` syntax.
+ * The `env()` helper is for TypeScript configuration files only.
+ *
+ * @see docs/design-docs/architecture-overhaul/typescript-configuration.md
+ */
+/**
+ * Read an environment variable, with an optional default.
+ *
+ * @param name - The environment variable name (e.g., `"SANITY_API_TOKEN"`)
+ * @param defaultValue - Fallback value if the variable is not set.
+ *   When omitted, a missing variable throws an error.
+ * @returns The environment variable's value, or the default.
+ * @throws {Error} When the variable is not set and no default is provided.
+ */
+export declare function env(name: string, defaultValue?: string): string;

package/dist/_vendor/ailf-core/env-helper.js

@@ -0,0 +1,45 @@
+/**
+ * env() — Typed environment variable access for configuration files.
+ *
+ * Provides a statically-analyzable alternative to raw `process.env` access.
+ * Tooling can detect which environment variables a config file depends on
+ * by scanning for `env("VAR_NAME")` calls.
+ *
+ * Usage:
+ * ```typescript
+ * import { env, defineConfig } from "@sanity/ailf"
+ *
+ * export default defineConfig({
+ *   source: {
+ *     projectId: env("SANITY_PROJECT_ID", "3do82whm"),
+ *     dataset: env("SANITY_DATASET", "next"),
+ *     apiToken: env("SANITY_API_TOKEN"), // throws if missing
+ *   },
+ * })
+ * ```
+ *
+ * YAML configs continue using the existing `${{ VAR | default }}` syntax.
+ * The `env()` helper is for TypeScript configuration files only.
+ *
+ * @see docs/design-docs/architecture-overhaul/typescript-configuration.md
+ */
+/**
+ * Read an environment variable, with an optional default.
+ *
+ * @param name - The environment variable name (e.g., `"SANITY_API_TOKEN"`)
+ * @param defaultValue - Fallback value if the variable is not set.
+ *   When omitted, a missing variable throws an error.
+ * @returns The environment variable's value, or the default.
+ * @throws {Error} When the variable is not set and no default is provided.
+ */
+export function env(name, defaultValue) {
+    const value = process.env[name];
+    if (value !== undefined && value !== "") {
+        return value;
+    }
+    if (defaultValue !== undefined) {
+        return defaultValue;
+    }
+    throw new Error(`Missing required environment variable: ${name}\n` +
+        `Set it in your shell, .env file, or provide a default: env("${name}", "fallback")`);
+}

package/dist/_vendor/ailf-core/examples/index.d.ts

@@ -140,12 +140,12 @@ export declare const exampleGroqBlogListingData: readonly [{
     }];
     readonly baseline: {
         readonly enabled: true;
-        readonly rubric: "abbreviated";
+        readonly rubric: "full";
     };
     readonly status: "draft";
 }];
 /** Raw YAML string for example-groq-blog-listing (preserves comments) */
-export declare const exampleGroqBlogListingYaml = "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n# Example Task: Blog listing with GROQ queries\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n#\n# This is a starter template \u2014 edit it for your own documentation.\n# Each task evaluates whether an AI coding agent can implement a feature\n# using your docs as context. Delete this file or replace it entirely.\n#\n# This example task ships as a DRAFT so it does not run in production\n#   evaluations automatically. To activate it, change status to \"active\"\n# or remove the status line entirely (defaults to active).\n#\n# Full field reference:\n#   https://github.com/sanity-labs/ai-literacy-framework/blob/main/docs/CONTRIBUTING_TASKS.md\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n\n# Unique identifier \u2014 lowercase alphanumeric with hyphens.\n# Must be unique across all task files in .ailf/tasks/.\n- id: example-groq-blog-listing\n\n  # Short human-readable summary. Shown in score tables and reports.\n  description: \"Example \u2014 Blog listing with GROQ queries\"\n\n  # Feature area this task belongs to. Tasks with the same area are\n  # grouped together in score summaries. Use a short kebab-case name.\n  featureArea: groq\n\n  # Gold-standard documentation articles for this task. The pipeline\n  # fetches these from Sanity and injects them into the prompt for\n  # baseline evaluation. Each entry needs:\n  #   slug   \u2014 the article's URL slug in your docs site\n  #   reason \u2014 why this doc is relevant (helps with auditing)\n  #\n  # This example uses slug-based references \u2014 the simplest form.\n  # See the other example tasks for path, id, and perspective references.\n  canonicalDocs:\n    - slug: groq-introduction\n      reason: \"Core GROQ syntax and query language reference\"\n    - slug: how-queries-work\n      reason: \"Query execution model and best practices\"\n\n  # When true, the pipeline auto-generates an additional rubric that\n  # checks whether the LLM's response actually used the provided docs.\n  docCoverage: true\n\n  # Path to a gold-standard implementation, relative to canonical/.\n  # The grader uses this as a reference when scoring code correctness.\n  referenceSolution: canonical/example-groq-blog-listing.ts\n\n  # vars.task \u2014 the implementation prompt given to the LLM.\n  # Write this as if you're asking a developer to build the feature.\n  # Be specific about requirements so the grader can evaluate clearly.\n  #\n  # vars.docs \u2014 leave empty (\"\"). The pipeline fills this in:\n  #   \u2022 Gold variant: injected with canonical doc content\n  #   \u2022 Baseline variant: left empty (tests model knowledge alone)\n  vars:\n    task: |\n      Create a Next.js page component that lists blog posts from Sanity\n      using GROQ. The page should display the title, slug, and published\n      date for each post, sorted by most recent first. Use the Sanity\n      client to fetch data.\n    docs: \"\"\n\n  # Grading assertions \u2014 how the LLM's response is scored.\n  #\n  # \"llm-rubric\" assertions use a grader LLM to score against criteria.\n  # The \"template\" references a rubric from config/rubrics.yaml.\n  # The \"criteria\" are task-specific bullets injected into the template.\n  #\n  # Available templates:\n  #   task-completion   \u2014 did the LLM implement the feature? (weight: 0.50)\n  #   code-correctness  \u2014 is the code idiomatic and correct? (weight: 0.25)\n  #\n  # You can also use value-based assertions:\n  #   - type: contains\n  #     value: \"client.fetch\"\n  #   - type: contains-any\n  #     value: [\"createClient\", \"sanityClient\"]\n  assert:\n    - type: llm-rubric\n      template: task-completion\n      criteria:\n        - \"Uses the groq tagged template literal\"\n        - \"Fetches blog posts with title, slug, and publishedAt fields\"\n        - \"Orders results by publishedAt in descending order\"\n\n    - type: llm-rubric\n      template: code-correctness\n      criteria:\n        - \"Uses createClient from @sanity/client or next-sanity\"\n        - \"Exports a valid Next.js page component\"\n\n  # Baseline variant configuration.\n  #   enabled \u2014 set to false to skip this task entirely\n  #   rubric  \u2014 \"abbreviated\" (faster, default), \"full\", or \"none\"\n  baseline:\n    enabled: true\n    rubric: abbreviated\n\n  # Example tasks ship as drafts so they don't run in production evals.\n  # Change to \"active\" (or remove this line) to activate.\n  status: draft\n";
+export declare const exampleGroqBlogListingYaml = "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n# Example Task: Blog listing with GROQ queries\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n#\n# This is a starter template \u2014 edit it for your own documentation.\n# Each task evaluates whether an AI coding agent can implement a feature\n# using your docs as context. Delete this file or replace it entirely.\n#\n# This example task ships as a DRAFT so it does not run in production\n#   evaluations automatically. To activate it, change status to \"active\"\n# or remove the status line entirely (defaults to active).\n#\n# Full field reference:\n#   https://github.com/sanity-labs/ai-literacy-framework/blob/main/docs/CONTRIBUTING_TASKS.md\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n\n# Unique identifier \u2014 lowercase alphanumeric with hyphens.\n# Must be unique across all task files in .ailf/tasks/.\n- id: example-groq-blog-listing\n\n  # Short human-readable summary. Shown in score tables and reports.\n  description: \"Example \u2014 Blog listing with GROQ queries\"\n\n  # Feature area this task belongs to. Tasks with the same area are\n  # grouped together in score summaries. Use a short kebab-case name.\n  featureArea: groq\n\n  # Gold-standard documentation articles for this task. The pipeline\n  # fetches these from Sanity and injects them into the prompt for\n  # baseline evaluation. Each entry needs:\n  #   slug   \u2014 the article's URL slug in your docs site\n  #   reason \u2014 why this doc is relevant (helps with auditing)\n  #\n  # This example uses slug-based references \u2014 the simplest form.\n  # See the other example tasks for path, id, and perspective references.\n  canonicalDocs:\n    - slug: groq-introduction\n      reason: \"Core GROQ syntax and query language reference\"\n    - slug: how-queries-work\n      reason: \"Query execution model and best practices\"\n\n  # When true, the pipeline auto-generates an additional rubric that\n  # checks whether the LLM's response actually used the provided docs.\n  docCoverage: true\n\n  # Path to a gold-standard implementation, relative to canonical/.\n  # The grader uses this as a reference when scoring code correctness.\n  referenceSolution: canonical/example-groq-blog-listing.ts\n\n  # vars.task \u2014 the implementation prompt given to the LLM.\n  # Write this as if you're asking a developer to build the feature.\n  # Be specific about requirements so the grader can evaluate clearly.\n  #\n  # vars.docs \u2014 leave empty (\"\"). The pipeline fills this in:\n  #   \u2022 Gold variant: injected with canonical doc content\n  #   \u2022 Baseline variant: left empty (tests model knowledge alone)\n  vars:\n    task: |\n      Create a Next.js page component that lists blog posts from Sanity\n      using GROQ. The page should display the title, slug, and published\n      date for each post, sorted by most recent first. Use the Sanity\n      client to fetch data.\n    docs: \"\"\n\n  # Grading assertions \u2014 how the LLM's response is scored.\n  #\n  # \"llm-rubric\" assertions use a grader LLM to score against criteria.\n  # The \"template\" references a rubric from config/rubrics.yaml.\n  # The \"criteria\" are task-specific bullets injected into the template.\n  #\n  # Available templates:\n  #   task-completion   \u2014 did the LLM implement the feature? (weight: 0.50)\n  #   code-correctness  \u2014 is the code idiomatic and correct? (weight: 0.25)\n  #\n  # You can also use value-based assertions:\n  #   - type: contains\n  #     value: \"client.fetch\"\n  #   - type: contains-any\n  #     value: [\"createClient\", \"sanityClient\"]\n  assert:\n    - type: llm-rubric\n      template: task-completion\n      criteria:\n        - \"Uses the groq tagged template literal\"\n        - \"Fetches blog posts with title, slug, and publishedAt fields\"\n        - \"Orders results by publishedAt in descending order\"\n\n    - type: llm-rubric\n      template: code-correctness\n      criteria:\n        - \"Uses createClient from @sanity/client or next-sanity\"\n        - \"Exports a valid Next.js page component\"\n\n  # Baseline variant configuration.\n  #   enabled \u2014 set to false to skip this task entirely\n  #   rubric  \u2014 \"full\" (default), \"abbreviated\" (faster), or \"none\"\n  baseline:\n    enabled: true\n    rubric: full\n\n  # Example tasks ship as drafts so they don't run in production evals.\n  # Change to \"active\" (or remove this line) to activate.\n  status: draft\n";
 /** Parsed task data for example-id-based-ref (JSON-safe) */
 export declare const exampleIdBasedRefData: readonly [{
     readonly id: "example-id-based-ref";
@@ -176,12 +176,12 @@ export declare const exampleIdBasedRefData: readonly [{
     }];
     readonly baseline: {
         readonly enabled: true;
-        readonly rubric: "abbreviated";
+        readonly rubric: "full";
     };
     readonly status: "draft";
 }];
 /** Raw YAML string for example-id-based-ref (preserves comments) */
-export declare const exampleIdBasedRefYaml = "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n# Example Task: Document ID-based canonical doc references\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n#\n# Demonstrates using `id` to reference canonical documentation by\n# Sanity document `_id`. This is useful for:\n#   - Draft documents that don't have a stable slug yet\n#   - Programmatic references from imports or migrations\n#   - Documents where you know the _id but not the slug\n#\n# The `id` ref type can also carry optional `slug` and `path` fields\n# as human-readable annotations \u2014 these are NOT used for resolution,\n# only for display in logs and reports.\n#\n# This example task ships as a DRAFT so it does not run in production\n#   evaluations automatically. To activate it, change status to \"active\"\n# or remove the status line entirely (defaults to active).\n#\n# @see docs/design-docs/canonical-doc-resolution.md\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n\n- id: example-id-based-ref\n  description: \"Example \u2014 GROQ feature support (ID-based doc references)\"\n\n  featureArea: groq\n\n  # ID-based canonical doc references.\n  #\n  # Use the Sanity document _id to reference articles directly.\n  # Optional slug/path annotations help humans reading the YAML\n  # but are NOT used for resolution \u2014 only the `id` field matters.\n  #\n  # These IDs reference real articles in the Sanity docs (next dataset):\n  #   0ba88f1b... = \"GROQ feature support across Sanity\"\n  #   5b9c2863... = \"Custom GROQ functions\"\n  canonicalDocs:\n    - id: \"0ba88f1b-d1a7-418a-9267-2e343d01886a\"\n      slug: groq-feature-support-by-context # annotation only \u2014 not used for resolution\n      reason: \"GROQ feature support across different Sanity contexts\"\n    - id: \"5b9c2863-ef01-4565-af8e-ee54e081ee74\"\n      slug: custom-groq-functions # annotation only \u2014 not used for resolution\n      reason: \"Custom GROQ functions and pipelines\"\n\n  docCoverage: true\n\n  vars:\n    task: |\n      Explain how GROQ is used across different Sanity contexts.\n      Cover the following:\n      1. Which GROQ features are available in each context (API queries,\n         webhooks, custom functions, access control)\n      2. How to create and use custom GROQ functions\n      3. Any differences in GROQ support between contexts\n      Provide examples demonstrating context-specific GROQ patterns.\n    docs: \"\"\n\n  assert:\n    - type: llm-rubric\n      template: task-completion\n      criteria:\n        - \"Explains GROQ availability across different Sanity contexts\"\n        - \"Describes custom GROQ function creation and usage\"\n        - \"Notes differences in GROQ support between contexts\"\n\n    - type: llm-rubric\n      template: code-correctness\n      criteria:\n        - \"GROQ examples use valid syntax\"\n        - \"Custom function examples follow the correct API pattern\"\n\n  baseline:\n    enabled: true\n    rubric: abbreviated\n\n  # Example tasks ship as drafts so they don't run in production evals.\n  # Change to \"active\" (or remove this line) to activate.\n  status: draft\n";
+export declare const exampleIdBasedRefYaml = "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n# Example Task: Document ID-based canonical doc references\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n#\n# Demonstrates using `id` to reference canonical documentation by\n# Sanity document `_id`. This is useful for:\n#   - Draft documents that don't have a stable slug yet\n#   - Programmatic references from imports or migrations\n#   - Documents where you know the _id but not the slug\n#\n# The `id` ref type can also carry optional `slug` and `path` fields\n# as human-readable annotations \u2014 these are NOT used for resolution,\n# only for display in logs and reports.\n#\n# This example task ships as a DRAFT so it does not run in production\n#   evaluations automatically. To activate it, change status to \"active\"\n# or remove the status line entirely (defaults to active).\n#\n# @see docs/design-docs/canonical-doc-resolution.md\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n\n- id: example-id-based-ref\n  description: \"Example \u2014 GROQ feature support (ID-based doc references)\"\n\n  featureArea: groq\n\n  # ID-based canonical doc references.\n  #\n  # Use the Sanity document _id to reference articles directly.\n  # Optional slug/path annotations help humans reading the YAML\n  # but are NOT used for resolution \u2014 only the `id` field matters.\n  #\n  # These IDs reference real articles in the Sanity docs (next dataset):\n  #   0ba88f1b... = \"GROQ feature support across Sanity\"\n  #   5b9c2863... = \"Custom GROQ functions\"\n  canonicalDocs:\n    - id: \"0ba88f1b-d1a7-418a-9267-2e343d01886a\"\n      slug: groq-feature-support-by-context # annotation only \u2014 not used for resolution\n      reason: \"GROQ feature support across different Sanity contexts\"\n    - id: \"5b9c2863-ef01-4565-af8e-ee54e081ee74\"\n      slug: custom-groq-functions # annotation only \u2014 not used for resolution\n      reason: \"Custom GROQ functions and pipelines\"\n\n  docCoverage: true\n\n  vars:\n    task: |\n      Explain how GROQ is used across different Sanity contexts.\n      Cover the following:\n      1. Which GROQ features are available in each context (API queries,\n         webhooks, custom functions, access control)\n      2. How to create and use custom GROQ functions\n      3. Any differences in GROQ support between contexts\n      Provide examples demonstrating context-specific GROQ patterns.\n    docs: \"\"\n\n  assert:\n    - type: llm-rubric\n      template: task-completion\n      criteria:\n        - \"Explains GROQ availability across different Sanity contexts\"\n        - \"Describes custom GROQ function creation and usage\"\n        - \"Notes differences in GROQ support between contexts\"\n\n    - type: llm-rubric\n      template: code-correctness\n      criteria:\n        - \"GROQ examples use valid syntax\"\n        - \"Custom function examples follow the correct API pattern\"\n\n  baseline:\n    enabled: true\n    rubric: full\n\n  # Example tasks ship as drafts so they don't run in production evals.\n  # Change to \"active\" (or remove this line) to activate.\n  status: draft\n";
 /** Parsed task data for example-path-based-ref (JSON-safe) */
 export declare const examplePathBasedRefData: readonly [{
     readonly id: "example-path-based-ref";
@@ -210,12 +210,12 @@ export declare const examplePathBasedRefData: readonly [{
     }];
     readonly baseline: {
         readonly enabled: true;
-        readonly rubric: "abbreviated";
+        readonly rubric: "full";
     };
     readonly status: "draft";
 }];
 /** Raw YAML string for example-path-based-ref (preserves comments) */
-export declare const examplePathBasedRefYaml = "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n# Example Task: Path-based canonical doc references\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n#\n# Demonstrates using `path` to reference canonical documentation.\n# Paths are the preferred reference type because they uniquely identify\n# an article across sections (unlike slugs, which can collide).\n#\n# Path format:\n#   - Simple:   \"webhooks\"              \u2192 resolves by slug lookup\n#   - Sectioned: \"content-lake/webhooks\" \u2192 disambiguates by section + slug\n#\n# This example demonstrates why paths matter: the slug \"documents\"\n# exists in both the \"content-lake\" and \"cli-reference\" sections.\n# Using \"content-lake/documents\" ensures we get the right one.\n#\n# This example task ships as a DRAFT so it does not run in production\n#   evaluations automatically. To activate it, change status to \"active\"\n# or remove the status line entirely (defaults to active).\n#\n# @see docs/design-docs/canonical-doc-resolution.md\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n\n- id: example-path-based-ref\n  description: \"Example \u2014 GROQ mutations (path-based doc references)\"\n\n  featureArea: groq\n\n  # Path-based canonical doc references.\n  #\n  # Use \"section/slug\" format to uniquely identify articles:\n  #   - \"content-lake/mutations-introduction\" \u2192 the mutations article\n  #   - \"content-lake/documents\" \u2192 the documents article in Content Lake\n  #     (not the CLI \"documents\" article in cli-reference section)\n  #\n  # The \"documents\" slug exists in two sections \u2014 this is exactly why\n  # path-based references are preferred over slug-based references.\n  canonicalDocs:\n    - path: content-lake/mutations-introduction\n      reason: \"Introduction to document mutations in the Content Lake\"\n    - path: content-lake/documents\n      reason: \"Document structure and types (Content Lake, not CLI reference)\"\n\n  docCoverage: true\n\n  vars:\n    task: |\n      Explain how to create, update, and delete documents in Sanity's\n      Content Lake using mutations. Cover:\n      1. The different mutation types (create, createOrReplace, patch, delete)\n      2. Document structure and required fields (_id, _type)\n      3. How to use patch operations to update specific fields\n      4. Best practices for mutation patterns\n      Provide working code examples using @sanity/client.\n    docs: \"\"\n\n  assert:\n    - type: llm-rubric\n      template: task-completion\n      criteria:\n        - \"Explains create, createOrReplace, patch, and delete mutations\"\n        - \"Describes required document fields (_id, _type)\"\n        - \"Shows patch operations for field-level updates\"\n        - \"Includes practical code examples\"\n\n    - type: llm-rubric\n      template: code-correctness\n      criteria:\n        - \"Uses correct @sanity/client mutation API\"\n        - \"Patch operations use valid set/unset/inc syntax\"\n\n  baseline:\n    enabled: true\n    rubric: abbreviated\n\n  # Example tasks ship as drafts so they don't run in production evals.\n  # Change to \"active\" (or remove this line) to activate.\n  status: draft\n";
+export declare const examplePathBasedRefYaml = "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n# Example Task: Path-based canonical doc references\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n#\n# Demonstrates using `path` to reference canonical documentation.\n# Paths are the preferred reference type because they uniquely identify\n# an article across sections (unlike slugs, which can collide).\n#\n# Path format:\n#   - Simple:   \"webhooks\"              \u2192 resolves by slug lookup\n#   - Sectioned: \"content-lake/webhooks\" \u2192 disambiguates by section + slug\n#\n# This example demonstrates why paths matter: the slug \"documents\"\n# exists in both the \"content-lake\" and \"cli-reference\" sections.\n# Using \"content-lake/documents\" ensures we get the right one.\n#\n# This example task ships as a DRAFT so it does not run in production\n#   evaluations automatically. To activate it, change status to \"active\"\n# or remove the status line entirely (defaults to active).\n#\n# @see docs/design-docs/canonical-doc-resolution.md\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n\n- id: example-path-based-ref\n  description: \"Example \u2014 GROQ mutations (path-based doc references)\"\n\n  featureArea: groq\n\n  # Path-based canonical doc references.\n  #\n  # Use \"section/slug\" format to uniquely identify articles:\n  #   - \"content-lake/mutations-introduction\" \u2192 the mutations article\n  #   - \"content-lake/documents\" \u2192 the documents article in Content Lake\n  #     (not the CLI \"documents\" article in cli-reference section)\n  #\n  # The \"documents\" slug exists in two sections \u2014 this is exactly why\n  # path-based references are preferred over slug-based references.\n  canonicalDocs:\n    - path: content-lake/mutations-introduction\n      reason: \"Introduction to document mutations in the Content Lake\"\n    - path: content-lake/documents\n      reason: \"Document structure and types (Content Lake, not CLI reference)\"\n\n  docCoverage: true\n\n  vars:\n    task: |\n      Explain how to create, update, and delete documents in Sanity's\n      Content Lake using mutations. Cover:\n      1. The different mutation types (create, createOrReplace, patch, delete)\n      2. Document structure and required fields (_id, _type)\n      3. How to use patch operations to update specific fields\n      4. Best practices for mutation patterns\n      Provide working code examples using @sanity/client.\n    docs: \"\"\n\n  assert:\n    - type: llm-rubric\n      template: task-completion\n      criteria:\n        - \"Explains create, createOrReplace, patch, and delete mutations\"\n        - \"Describes required document fields (_id, _type)\"\n        - \"Shows patch operations for field-level updates\"\n        - \"Includes practical code examples\"\n\n    - type: llm-rubric\n      template: code-correctness\n      criteria:\n        - \"Uses correct @sanity/client mutation API\"\n        - \"Patch operations use valid set/unset/inc syntax\"\n\n  baseline:\n    enabled: true\n    rubric: full\n\n  # Example tasks ship as drafts so they don't run in production evals.\n  # Change to \"active\" (or remove this line) to activate.\n  status: draft\n";
 /** Parsed task data for example-perspective-ref (JSON-safe) */
 export declare const examplePerspectiveRefData: readonly [{
     readonly id: "example-perspective-ref";
@@ -244,12 +244,12 @@ export declare const examplePerspectiveRefData: readonly [{
     }];
     readonly baseline: {
         readonly enabled: true;
-        readonly rubric: "abbreviated";
+        readonly rubric: "full";
     };
     readonly status: "draft";
 }];
 /** Raw YAML string for example-perspective-ref (preserves comments) */
-export declare const examplePerspectiveRefYaml = "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n# Example Task: Perspective / content release doc references\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n#\n# Demonstrates using `perspective` to reference all documentation\n# articles within a content release. This is the key capability for\n# evaluating NEW feature documentation before it's published.\n#\n# How it works:\n#   - A perspective ref is one-to-many: the doc fetcher queries the\n#     named release and expands it to ALL articles versioned within it.\n#   - Downstream consumers see the same flat DocContext[] regardless\n#     of how docs were resolved.\n#   - When the release is published, the perspective entry becomes a\n#     no-op (articles are now in published). Migrate to explicit path\n#     or slug refs at your convenience.\n#\n# This example task ships as a DRAFT so it does not run in production\n#   evaluations automatically. To activate it, change status to \"active\"\n# or remove the status line entirely (defaults to active).\n#\n# @see docs/design-docs/canonical-doc-resolution.md\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n\n- id: example-perspective-ref\n  description:\n    \"Example \u2014 GROQ features from content release (perspective-based doc\n    references)\"\n\n  featureArea: groq\n\n  # Perspective-based canonical doc reference.\n  #\n  # The perspective ID references a content release in the Sanity\n  # Content Lake. At evaluation time, the doc fetcher auto-discovers\n  # all articles versioned in this release and includes them as\n  # canonical documentation context.\n  #\n  # Release rE9TSJvR4 contains:\n  #   - \"GROQ-powered webhooks\" (webhooks)\n  #   - \"Query Cheat Sheet - GROQ\" (query-cheat-sheet)\n  #   - \"GROQ joins\" (groq-joins)\n  #\n  # You can combine perspective refs with explicit slug/path/id refs\n  # to include foundational published docs alongside release content.\n  # Here we add groq-data-types as a complementary published reference.\n  canonicalDocs:\n    - perspective: rE9TSJvR4\n      reason: \"All GROQ documentation updates in the test content release\"\n    - slug: groq-data-types\n      reason: \"GROQ data type reference (published, stable)\"\n\n  docCoverage: true\n\n  vars:\n    task: |\n      Using GROQ, demonstrate advanced query patterns including:\n      1. Joining data across document types using references\n      2. Filtering webhook payloads with GROQ projections\n      3. Using the query cheat sheet patterns for common operations\n      4. Working with different GROQ data types in filters\n      Provide working GROQ query examples for each pattern.\n    docs: \"\"\n\n  assert:\n    - type: llm-rubric\n      template: task-completion\n      criteria:\n        - \"Demonstrates GROQ join syntax for cross-document queries\"\n        - \"Shows GROQ filter patterns for webhook configuration\"\n        - \"Includes practical query examples from cheat sheet patterns\"\n\n    - type: llm-rubric\n      template: code-correctness\n      criteria:\n        - \"All GROQ queries use valid syntax\"\n        - \"Reference joins use correct dereference operator (->)\"\n\n  baseline:\n    enabled: true\n    rubric: abbreviated\n\n  # Example tasks ship as drafts so they don't run in production evals.\n  # Change to \"active\" (or remove this line) to activate.\n  status: draft\n";
+export declare const examplePerspectiveRefYaml = "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n# Example Task: Perspective / content release doc references\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n#\n# Demonstrates using `perspective` to reference all documentation\n# articles within a content release. This is the key capability for\n# evaluating NEW feature documentation before it's published.\n#\n# How it works:\n#   - A perspective ref is one-to-many: the doc fetcher queries the\n#     named release and expands it to ALL articles versioned within it.\n#   - Downstream consumers see the same flat DocContext[] regardless\n#     of how docs were resolved.\n#   - When the release is published, the perspective entry becomes a\n#     no-op (articles are now in published). Migrate to explicit path\n#     or slug refs at your convenience.\n#\n# This example task ships as a DRAFT so it does not run in production\n#   evaluations automatically. To activate it, change status to \"active\"\n# or remove the status line entirely (defaults to active).\n#\n# @see docs/design-docs/canonical-doc-resolution.md\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n\n- id: example-perspective-ref\n  description:\n    \"Example \u2014 GROQ features from content release (perspective-based doc\n    references)\"\n\n  featureArea: groq\n\n  # Perspective-based canonical doc reference.\n  #\n  # The perspective ID references a content release in the Sanity\n  # Content Lake. At evaluation time, the doc fetcher auto-discovers\n  # all articles versioned in this release and includes them as\n  # canonical documentation context.\n  #\n  # Release rE9TSJvR4 contains:\n  #   - \"GROQ-powered webhooks\" (webhooks)\n  #   - \"Query Cheat Sheet - GROQ\" (query-cheat-sheet)\n  #   - \"GROQ joins\" (groq-joins)\n  #\n  # You can combine perspective refs with explicit slug/path/id refs\n  # to include foundational published docs alongside release content.\n  # Here we add groq-data-types as a complementary published reference.\n  canonicalDocs:\n    - perspective: rE9TSJvR4\n      reason: \"All GROQ documentation updates in the test content release\"\n    - slug: groq-data-types\n      reason: \"GROQ data type reference (published, stable)\"\n\n  docCoverage: true\n\n  vars:\n    task: |\n      Using GROQ, demonstrate advanced query patterns including:\n      1. Joining data across document types using references\n      2. Filtering webhook payloads with GROQ projections\n      3. Using the query cheat sheet patterns for common operations\n      4. Working with different GROQ data types in filters\n      Provide working GROQ query examples for each pattern.\n    docs: \"\"\n\n  assert:\n    - type: llm-rubric\n      template: task-completion\n      criteria:\n        - \"Demonstrates GROQ join syntax for cross-document queries\"\n        - \"Shows GROQ filter patterns for webhook configuration\"\n        - \"Includes practical query examples from cheat sheet patterns\"\n\n    - type: llm-rubric\n      template: code-correctness\n      criteria:\n        - \"All GROQ queries use valid syntax\"\n        - \"Reference joins use correct dereference operator (->)\"\n\n  baseline:\n    enabled: true\n    rubric: full\n\n  # Example tasks ship as drafts so they don't run in production evals.\n  # Change to \"active\" (or remove this line) to activate.\n  status: draft\n";
 /** Parsed task data for example-studio-custom-input (JSON-safe) */
 export declare const exampleStudioCustomInputData: readonly [{
     readonly id: "example-studio-custom-input";
@@ -279,12 +279,12 @@ export declare const exampleStudioCustomInputData: readonly [{
     }];
     readonly baseline: {
         readonly enabled: true;
-        readonly rubric: "abbreviated";
+        readonly rubric: "full";
     };
     readonly status: "draft";
 }];
 /** Raw YAML string for example-studio-custom-input (preserves comments) */
-export declare const exampleStudioCustomInputYaml = "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n# Example Task: Custom input component in Sanity Studio\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n#\n# This is a starter template \u2014 edit it for your own documentation.\n# Delete this file or replace it with your own tasks.\n#\n# This example task ships as a DRAFT so it does not run in production\n#   evaluations automatically. To activate it, change status to \"active\"\n# or remove the status line entirely (defaults to active).\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n\n- id: example-studio-custom-input\n  description: \"Example \u2014 Custom input component in Sanity Studio\"\n\n  featureArea: studio\n\n  # Slug-based canonical doc references.\n  canonicalDocs:\n    - slug: custom-input-widgets\n      reason: \"Guide for building custom form inputs in Sanity Studio\"\n    - slug: form-components\n      reason: \"Form component API and customization patterns\"\n\n  docCoverage: true\n  referenceSolution: canonical/example-studio-custom-input.ts\n\n  vars:\n    task: |\n      Build a custom string input component for Sanity Studio that shows\n      a character count below the input field. The component should accept\n      a maxLength option from the field schema and display a warning when\n      the text exceeds the limit.\n    docs: \"\"\n\n  assert:\n    - type: llm-rubric\n      template: task-completion\n      criteria:\n        - \"Implements a React component that renders a text input\"\n        - \"Displays a live character count\"\n        - \"Reads maxLength from schema options\"\n        - \"Shows a visual warning when limit is exceeded\"\n\n    - type: llm-rubric\n      template: code-correctness\n      criteria:\n        - \"Uses the Sanity UI library for styling\"\n        - \"Calls onChange with patch operations\"\n\n  baseline:\n    enabled: true\n    rubric: abbreviated\n\n  # Example tasks ship as drafts so they don't run in production evals.\n  # Change to \"active\" (or remove this line) to activate.\n  status: draft\n";
+export declare const exampleStudioCustomInputYaml = "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n# Example Task: Custom input component in Sanity Studio\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n#\n# This is a starter template \u2014 edit it for your own documentation.\n# Delete this file or replace it with your own tasks.\n#\n# This example task ships as a DRAFT so it does not run in production\n#   evaluations automatically. To activate it, change status to \"active\"\n# or remove the status line entirely (defaults to active).\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n\n- id: example-studio-custom-input\n  description: \"Example \u2014 Custom input component in Sanity Studio\"\n\n  featureArea: studio\n\n  # Slug-based canonical doc references.\n  canonicalDocs:\n    - slug: custom-input-widgets\n      reason: \"Guide for building custom form inputs in Sanity Studio\"\n    - slug: form-components\n      reason: \"Form component API and customization patterns\"\n\n  docCoverage: true\n  referenceSolution: canonical/example-studio-custom-input.ts\n\n  vars:\n    task: |\n      Build a custom string input component for Sanity Studio that shows\n      a character count below the input field. The component should accept\n      a maxLength option from the field schema and display a warning when\n      the text exceeds the limit.\n    docs: \"\"\n\n  assert:\n    - type: llm-rubric\n      template: task-completion\n      criteria:\n        - \"Implements a React component that renders a text input\"\n        - \"Displays a live character count\"\n        - \"Reads maxLength from schema options\"\n        - \"Shows a visual warning when limit is exceeded\"\n\n    - type: llm-rubric\n      template: code-correctness\n      criteria:\n        - \"Uses the Sanity UI library for styling\"\n        - \"Calls onChange with patch operations\"\n\n  baseline:\n    enabled: true\n    rubric: full\n\n  # Example tasks ship as drafts so they don't run in production evals.\n  # Change to \"active\" (or remove this line) to activate.\n  status: draft\n";
 /** All task example data as a flat array (JSON-safe) */
 export declare const allTaskData: readonly unknown[];
 /** Map of task ID (filename stem) → raw YAML string (preserves comments) */