@sanity/ailf 2.0.2 → 2.2.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@sanity/ailf",
-  "version": "2.0.2",
+  "version": "2.2.0",
   "private": false,
   "publishConfig": {
     "access": "public"
@@ -31,6 +31,29 @@
     "canonical",
     "tasks"
   ],
+  "dependencies": {
+    "@google-cloud/bigquery": "^8.1.1",
+    "@inquirer/prompts": "^8.3.0",
+    "@modelcontextprotocol/sdk": "^1.29.0",
+    "@portabletext/markdown": "^1.0.0",
+    "@sanity/client": "^7.3.0",
+    "commander": "^14.0.3",
+    "dotenv": "^16.4.7",
+    "dotenv-cli": "^11.0.0",
+    "jiti": "^2.6.1",
+    "js-yaml": "^4.1.0",
+    "promptfoo": "^0.120.24",
+    "zod": "^4.3.6"
+  },
+  "devDependencies": {
+    "@anthropic-ai/claude-agent-sdk": "^0.2.105",
+    "@types/js-yaml": "^4.0.9",
+    "@types/node": "^22.13.1",
+    "tsx": "^4.19.2",
+    "typescript": "^5.7.3",
+    "@sanity/ailf-core": "0.1.0",
+    "@sanity/ailf-shared": "0.1.0"
+  },
   "scripts": {
     "build": "tsc && tsx scripts/bundle-workspace-deps.ts",
     "generate-configs": "tsx src/cli.ts generate-configs",
@@ -58,27 +81,5 @@
     "discovery-report": "tsx src/cli.ts discovery-report",
     "webhook-server": "tsx src/cli.ts webhook-server",
     "weekly-digest": "tsx src/cli.ts weekly-digest"
-  },
-  "dependencies": {
-    "@google-cloud/bigquery": "^8.1.1",
-    "@inquirer/prompts": "^8.3.0",
-    "@modelcontextprotocol/sdk": "^1.29.0",
-    "@portabletext/markdown": "^1.0.0",
-    "@sanity/client": "^7.3.0",
-    "commander": "^14.0.3",
-    "dotenv": "^16.4.7",
-    "dotenv-cli": "^11.0.0",
-    "jiti": "^2.6.1",
-    "js-yaml": "^4.1.0",
-    "promptfoo": "^0.120.24",
-    "zod": "^4.3.6"
-  },
-  "devDependencies": {
-    "@sanity/ailf-core": "workspace:*",
-    "@sanity/ailf-shared": "workspace:*",
-    "@types/js-yaml": "^4.0.9",
-    "@types/node": "^22.13.1",
-    "tsx": "^4.19.2",
-    "typescript": "^5.7.3"
   }
-}
+}

package/dist/_vendor/ailf-core/__tests__/comparison-formatters.test.d.ts

@@ -1,10 +0,0 @@
-/**
- * comparison-formatters.test.ts
- *
- * Verifies that formatComparisonMarkdown() and formatComparisonTable()
- * dynamically derive column headers from the dimension keys present
- * in the report data, rather than hardcoding literacy-specific names.
- *
- * Run: npx tsx --test src/__tests__/comparison-formatters.test.ts
- */
-export {};

package/dist/_vendor/ailf-core/__tests__/comparison-formatters.test.js

@@ -1,185 +0,0 @@
-/**
- * comparison-formatters.test.ts
- *
- * Verifies that formatComparisonMarkdown() and formatComparisonTable()
- * dynamically derive column headers from the dimension keys present
- * in the report data, rather than hardcoding literacy-specific names.
- *
- * Run: npx tsx --test src/__tests__/comparison-formatters.test.ts
- */
-import assert from "node:assert/strict";
-import { describe, it } from "node:test";
-import { formatComparisonMarkdown, formatComparisonTable, } from "../services/comparison-formatters.js";
-// ---------------------------------------------------------------------------
-// Helpers
-// ---------------------------------------------------------------------------
-/** Minimal ScoreSummary stub — only fields the formatters actually read */
-function stubSummary(avgScore) {
-    return {
-        belowCritical: [],
-        lowestArea: "area-a",
-        lowestScore: 40,
-        overall: {
-            avgCeilingScore: 80,
-            avgScore,
-            avgDocLift: 10,
-            avgDocQualityGap: 20,
-            avgFloorScore: 30,
-            negativeDocLiftCount: 0,
-        },
-        scores: [],
-        timestamp: "2026-04-05T00:00:00.000Z",
-    };
-}
-function makeReport(overrides) {
-    return {
-        areas: [
-            {
-                area: "area-a",
-                baseline: 60,
-                experiment: 65,
-                delta: 5,
-                change: "improved",
-                dimensions: overrides.areaDimensions,
-                ceilingDelta: 0,
-                docLiftDelta: 2,
-                floorDelta: 0,
-            },
-        ],
-        baseline: stubSummary(60),
-        experiment: stubSummary(65),
-        deltas: {
-            overall: 5,
-            perArea: { "area-a": 5 },
-            perDimension: overrides.perDimension,
-            docLift: 2,
-        },
-        generatedAt: "2026-04-05T00:00:00.000Z",
-        improved: ["area-a"],
-        regressed: [],
-        unchanged: [],
-        notEvaluated: [],
-        mismatched: { onlyInBaseline: [], onlyInExperiment: [] },
-        noiseThreshold: 2,
-        noiseThresholdEmpirical: false,
-    };
-}
-// ---------------------------------------------------------------------------
-// Tests — literacy dimensions (backward compatibility)
-// ---------------------------------------------------------------------------
-describe("formatComparisonMarkdown", () => {
-    it("renders literacy dimension columns dynamically", () => {
-        const report = makeReport({
-            areaDimensions: {
-                "task-completion": { baseline: 60, experiment: 65, delta: 5 },
-                "code-correctness": { baseline: 50, experiment: 55, delta: 5 },
-                "doc-coverage": { baseline: 40, experiment: 42, delta: 2 },
-            },
-            perDimension: {
-                "task-completion": 5,
-                "code-correctness": 5,
-                "doc-coverage": 2,
-            },
-        });
-        const md = formatComparisonMarkdown(report);
-        // Column headers should be title-cased from kebab-case
-        assert.ok(md.includes("Task Completion"), "should have Task Completion column header");
-        assert.ok(md.includes("Code Correctness"), "should have Code Correctness column header");
-        assert.ok(md.includes("Doc Coverage"), "should have Doc Coverage column header");
-        // Per-dimension averages section should also show dynamic labels
-        assert.ok(md.includes("| Task Completion |"), "dimension averages should include Task Completion");
-        assert.ok(md.includes("| Code Correctness |"), "dimension averages should include Code Correctness");
-        assert.ok(md.includes("| Doc Coverage |"), "dimension averages should include Doc Coverage");
-    });
-    it("renders MCP dimension columns dynamically", () => {
-        const report = makeReport({
-            areaDimensions: {
-                "input-validation": { baseline: 50, experiment: 60, delta: 10 },
-                "output-correctness": { baseline: 70, experiment: 75, delta: 5 },
-                "error-handling": { baseline: 40, experiment: 45, delta: 5 },
-                security: { baseline: 80, experiment: 82, delta: 2 },
-            },
-            perDimension: {
-                "input-validation": 10,
-                "output-correctness": 5,
-                "error-handling": 5,
-                security: 2,
-            },
-        });
-        const md = formatComparisonMarkdown(report);
-        // 4 MCP columns instead of 3 literacy columns
-        assert.ok(md.includes("Input Validation"), "should have Input Validation column");
-        assert.ok(md.includes("Output Correctness"), "should have Output Correctness column");
-        assert.ok(md.includes("Error Handling"), "should have Error Handling column");
-        assert.ok(md.includes("Security"), "should have Security column");
-        // Per-dimension averages
-        assert.ok(md.includes("| Input Validation |"), "dimension averages should include Input Validation");
-        assert.ok(md.includes("| Security |"), "dimension averages should include Security");
-    });
-});
-describe("formatComparisonTable", () => {
-    it("renders literacy dimension columns dynamically", () => {
-        const report = makeReport({
-            areaDimensions: {
-                "task-completion": { baseline: 60, experiment: 65, delta: 5 },
-                "code-correctness": { baseline: 50, experiment: 55, delta: 5 },
-                "doc-coverage": { baseline: 40, experiment: 42, delta: 2 },
-            },
-            perDimension: {
-                "task-completion": 5,
-                "code-correctness": 5,
-                "doc-coverage": 2,
-            },
-        });
-        const table = formatComparisonTable(report);
-        // Dimension averages section
-        assert.ok(table.includes("Task Completion:"), "should show Task Completion in dimension averages");
-        assert.ok(table.includes("Code Correctness:"), "should show Code Correctness in dimension averages");
-        assert.ok(table.includes("Doc Coverage:"), "should show Doc Coverage in dimension averages");
-        // Per-area table header
-        assert.ok(table.includes("Task Completion"), "per-area table should have Task Completion header");
-        assert.ok(table.includes("Code Correctness"), "per-area table should have Code Correctness header");
-        assert.ok(table.includes("Doc Coverage"), "per-area table should have Doc Coverage header");
-    });
-    it("renders MCP dimension columns dynamically", () => {
-        const report = makeReport({
-            areaDimensions: {
-                "input-validation": { baseline: 50, experiment: 60, delta: 10 },
-                "output-correctness": { baseline: 70, experiment: 75, delta: 5 },
-                "error-handling": { baseline: 40, experiment: 45, delta: 5 },
-                security: { baseline: 80, experiment: 82, delta: 2 },
-            },
-            perDimension: {
-                "input-validation": 10,
-                "output-correctness": 5,
-                "error-handling": 5,
-                security: 2,
-            },
-        });
-        const table = formatComparisonTable(report);
-        // 4 MCP columns in the per-area table
-        assert.ok(table.includes("Input Validation"), "should have Input Validation");
-        assert.ok(table.includes("Output Correctness"), "should have Output Correctness");
-        assert.ok(table.includes("Error Handling"), "should have Error Handling");
-        assert.ok(table.includes("Security"), "should have Security");
-        // Should NOT have literacy dimension headers
-        assert.ok(!table.includes("Task Completion"), "should not contain Task Completion");
-        assert.ok(!table.includes("Doc Coverage"), "should not contain Doc Coverage");
-    });
-    it("includes delta values for each dimension in the per-area rows", () => {
-        const report = makeReport({
-            areaDimensions: {
-                "input-validation": { baseline: 50, experiment: 60, delta: 10 },
-                "output-correctness": { baseline: 70, experiment: 75, delta: 5 },
-            },
-            perDimension: {
-                "input-validation": 10,
-                "output-correctness": 5,
-            },
-        });
-        const table = formatComparisonTable(report);
-        // The per-area row should include the delta values (+10 and +5)
-        assert.ok(table.includes("+10"), "should show +10 delta for area-a");
-        assert.ok(table.includes("+5"), "should show +5 delta for area-a");
-    });
-});

package/dist/_vendor/ailf-core/artifact-capture/__tests__/noop-collector.test.d.ts

@@ -1,6 +0,0 @@
-/**
- * noop-collector.test.ts — verifies the NoOpArtifactCollector is truly zero-cost.
- *
- * Run: npx tsx --test src/artifact-capture/__tests__/noop-collector.test.ts
- */
-export {};

package/dist/_vendor/ailf-core/artifact-capture/__tests__/noop-collector.test.js

@@ -1,42 +0,0 @@
-/**
- * noop-collector.test.ts — verifies the NoOpArtifactCollector is truly zero-cost.
- *
- * Run: npx tsx --test src/artifact-capture/__tests__/noop-collector.test.ts
- */
-import assert from "node:assert/strict";
-import { describe, it } from "node:test";
-import { NoOpArtifactCollector } from "../noop-collector.js";
-describe("NoOpArtifactCollector", () => {
-    it("enabled returns false", () => {
-        const collector = new NoOpArtifactCollector();
-        assert.equal(collector.enabled, false);
-    });
-    it("extrasEnabled returns false", () => {
-        const collector = new NoOpArtifactCollector();
-        assert.equal(collector.extrasEnabled, false);
-    });
-    it("capture() is callable and returns void", () => {
-        const collector = new NoOpArtifactCollector();
-        const result = collector.capture("step", "type", { data: true });
-        assert.equal(result, undefined);
-    });
-    it("captureFile() is callable and returns void", () => {
-        const collector = new NoOpArtifactCollector();
-        const result = collector.captureFile("step", "type", "/some/path");
-        assert.equal(result, undefined);
-    });
-    it("flush() returns zero-count result", async () => {
-        const collector = new NoOpArtifactCollector();
-        const result = await collector.flush();
-        assert.equal(result.artifactCount, 0);
-        assert.equal(result.destination, "");
-        assert.equal(result.totalBytes, 0);
-        assert.equal(result.compressed, false);
-    });
-    it("flush() returns the same frozen object every time", async () => {
-        const collector = new NoOpArtifactCollector();
-        const a = await collector.flush();
-        const b = await collector.flush();
-        assert.equal(a, b);
-    });
-});

package/dist/_vendor/ailf-tasks/cli.d.ts

@@ -1,8 +0,0 @@
-/**
- * cli.ts — Minimal CLI for standalone task validation.
- *
- * Usage:
- *   npx @sanity/ailf-tasks validate .ailf/tasks/
- *   npx @sanity/ailf-tasks validate             # defaults to .ailf/tasks/
- */
-export declare function run(): void;

package/dist/_vendor/ailf-tasks/cli.js

@@ -1,61 +0,0 @@
-/**
- * cli.ts — Minimal CLI for standalone task validation.
- *
- * Usage:
- *   npx @sanity/ailf-tasks validate .ailf/tasks/
- *   npx @sanity/ailf-tasks validate             # defaults to .ailf/tasks/
- */
-import { loadTaskDir } from "./parser.js";
-import { formatValidationResult, validateRepoTasks } from "./validation.js";
-export function run() {
-    const args = process.argv.slice(2);
-    const command = args[0];
-    if (command === "validate") {
-        const dir = args[1] ?? ".ailf/tasks";
-        validateCommand(dir);
-    }
-    else if (command === "--help" ||
-        command === "-h" ||
-        command === undefined) {
-        printUsage();
-    }
-    else {
-        console.error(`Unknown command: ${command}`);
-        printUsage();
-        process.exit(1);
-    }
-}
-function validateCommand(dir) {
-    try {
-        const tasks = loadTaskDir(dir);
-        // Run semantic validation
-        const result = validateRepoTasks(tasks);
-        const formatted = formatValidationResult(result);
-        console.log(`✅ ${tasks.length} task(s) validated from ${dir}`);
-        for (const task of tasks) {
-            console.log(`   ${task.id} — ${task.description}`);
-        }
-        if (result.warnings.length > 0 || result.errors.length > 0) {
-            console.log("");
-            console.log(formatted);
-        }
-        if (!result.valid) {
-            process.exit(1);
-        }
-    }
-    catch (err) {
-        console.error(`❌ ${err instanceof Error ? err.message : String(err)}`);
-        process.exit(1);
-    }
-}
-function printUsage() {
-    console.log("Usage: ailf-tasks <command> [options]");
-    console.log("");
-    console.log("Commands:");
-    console.log("  validate [dir]  Validate task YAML files (default: .ailf/tasks/)");
-    console.log("");
-    console.log("Examples:");
-    console.log("  ailf-tasks validate");
-    console.log("  ailf-tasks validate .ailf/tasks/");
-    console.log("  ailf-tasks validate /path/to/tasks/");
-}

package/dist/_vendor/ailf-tasks/index.d.ts

@@ -1,13 +0,0 @@
-/**
- * @sanity/ailf-tasks — Task definition schemas and YAML parser.
- *
- * Lightweight package for parsing and validating .ailf/tasks/*.yaml files
- * without depending on the full AILF CLI or its heavyweight dependencies
- * (Promptfoo, LLM SDKs, Sanity client).
- *
- * Usage:
- *   import { parseTaskFile, loadTaskDir, RepoTaskSchema } from '@sanity/ailf-tasks'
- */
-export { CURATED_ASSERTION_TYPES, RepoTaskFileSchema, RepoTaskSchema, RUBRIC_TEMPLATE_NAMES, type CuratedAssertionType, type RepoTask, type RubricTemplateName, } from "./schemas.js";
-export { loadTaskDir, parseTaskFile } from "./parser.js";
-export { detectSnakeCaseFields, formatValidationResult, validateRepoTasks, type ValidationMessage, type ValidationResult, } from "./validation.js";

package/dist/_vendor/ailf-tasks/index.js

@@ -1,16 +0,0 @@
-/**
- * @sanity/ailf-tasks — Task definition schemas and YAML parser.
- *
- * Lightweight package for parsing and validating .ailf/tasks/*.yaml files
- * without depending on the full AILF CLI or its heavyweight dependencies
- * (Promptfoo, LLM SDKs, Sanity client).
- *
- * Usage:
- *   import { parseTaskFile, loadTaskDir, RepoTaskSchema } from '@sanity/ailf-tasks'
- */
-// Schemas and types
-export { CURATED_ASSERTION_TYPES, RepoTaskFileSchema, RepoTaskSchema, RUBRIC_TEMPLATE_NAMES, } from "./schemas.js";
-// Parsing
-export { loadTaskDir, parseTaskFile } from "./parser.js";
-// Validation
-export { detectSnakeCaseFields, formatValidationResult, validateRepoTasks, } from "./validation.js";

package/dist/_vendor/ailf-tasks/parser.d.ts

@@ -1,27 +0,0 @@
-/**
- * parser.ts — Standalone task file and directory parsing.
- *
- * High-level functions for loading and validating .ailf/tasks/ YAML
- * files without any dependency on the eval pipeline.
- *
- * Usage:
- *   import { parseTaskFile, loadTaskDir } from '@sanity/ailf-tasks'
- */
-import { type RepoTask } from "./schemas.js";
-/**
- * Parse a single task YAML string and return validated tasks.
- *
- * @param content - Raw YAML string content
- * @param filename - Source filename (for error messages)
- * @returns Validated array of RepoTask objects
- * @throws Error if YAML parsing or Zod validation fails
- */
-export declare function parseTaskFile(content: string, filename?: string): RepoTask[];
-/**
- * Load and parse all task YAML files from a directory.
- *
- * @param dirPath - Path to directory containing .yaml/.yml files
- * @returns All validated tasks, sorted by filename
- * @throws Error if directory not found, no YAML files, or validation fails
- */
-export declare function loadTaskDir(dirPath: string): RepoTask[];

package/dist/_vendor/ailf-tasks/parser.js

@@ -1,73 +0,0 @@
-/**
- * parser.ts — Standalone task file and directory parsing.
- *
- * High-level functions for loading and validating .ailf/tasks/ YAML
- * files without any dependency on the eval pipeline.
- *
- * Usage:
- *   import { parseTaskFile, loadTaskDir } from '@sanity/ailf-tasks'
- */
-import { existsSync, readdirSync, readFileSync } from "fs";
-import { resolve } from "path";
-import { load } from "js-yaml";
-import { RepoTaskFileSchema } from "./schemas.js";
-// ---------------------------------------------------------------------------
-// Public API
-// ---------------------------------------------------------------------------
-/**
- * Parse a single task YAML string and return validated tasks.
- *
- * @param content - Raw YAML string content
- * @param filename - Source filename (for error messages)
- * @returns Validated array of RepoTask objects
- * @throws Error if YAML parsing or Zod validation fails
- */
-export function parseTaskFile(content, filename = "<string>") {
-    const parsed = load(content);
-    if (!Array.isArray(parsed)) {
-        throw new Error(`${filename} did not parse to an array of tasks. ` +
-            "Task files must contain a YAML array of task definitions.");
-    }
-    const result = RepoTaskFileSchema.safeParse(parsed);
-    if (!result.success) {
-        const messages = result.error.issues
-            .map((i) => `  [${i.path.join(".")}]: ${i.message}`)
-            .join("\n");
-        throw new Error(`Invalid task file "${filename}":\n${messages}`);
-    }
-    return result.data;
-}
-/**
- * Load and parse all task YAML files from a directory.
- *
- * @param dirPath - Path to directory containing .yaml/.yml files
- * @returns All validated tasks, sorted by filename
- * @throws Error if directory not found, no YAML files, or validation fails
- */
-export function loadTaskDir(dirPath) {
-    if (!existsSync(dirPath)) {
-        throw new Error(`Tasks directory not found: ${dirPath}\n` +
-            "  Expected a directory containing .ailf/tasks/*.yaml files.");
-    }
-    const yamlFiles = readdirSync(dirPath)
-        .filter((f) => (f.endsWith(".yaml") || f.endsWith(".yml")) && !f.startsWith("."))
-        .sort();
-    if (yamlFiles.length === 0) {
-        throw new Error(`No YAML files found in ${dirPath}\n` +
-            "  Expected .ailf/tasks/*.yaml files with task definitions.");
-    }
-    const allTasks = [];
-    for (const file of yamlFiles) {
-        const filePath = resolve(dirPath, file);
-        const content = readFileSync(filePath, "utf-8");
-        try {
-            const tasks = parseTaskFile(content, file);
-            allTasks.push(...tasks);
-        }
-        catch (err) {
-            const msg = err instanceof Error ? err.message : String(err);
-            throw new Error(`Failed to load ${file}:\n${msg}`, { cause: err });
-        }
-    }
-    return allTasks;
-}