* feat(assistant): optimize MCP and custom widget authoring * fix(assistant): isolate server-safe JSX analyzer * refactor(assistant): generate MCP tool staging * fix(assistant): preserve widget placement follow-up * fix(ci): allow fast gate to finish * fix(custom-widgets): address review findings * fix(custom-widgets): split preview evidence module * style: format assistant custom widget files * fix(custom-widgets): sync formatted skill schema
104 lines
4.3 KiB
TypeScript
104 lines
4.3 KiB
TypeScript
import { mkdir, writeFile } from "node:fs/promises";
|
|
import path from "node:path";
|
|
|
|
import { CUSTOM_WIDGET_AI_EVALUATION_CASES } from "./ai-evaluation-cases";
|
|
import { evaluateCustomWidgetAssistantCase } from "./ai-assistant-evaluation";
|
|
import type { CustomWidgetAssistantEvaluationResult } from "./ai-assistant-evaluation";
|
|
import { evaluateCustomWidgetCase, resolveAiEvaluationProviderConfig } from "./ai-evaluation";
|
|
import type { AiEvaluationResult } from "./ai-evaluation";
|
|
|
|
const { apiKey, baseUrl: providerBaseUrl, generatorModel, judgeModel } = resolveAiEvaluationProviderConfig(process.env);
|
|
if (!apiKey) {
|
|
throw new Error("AI_PROVIDER_API_KEY or OPENROUTER_API_KEY is required for the live Custom Widget AI evaluation");
|
|
}
|
|
|
|
const requestedCase = process.argv.find((value) => value.startsWith("--case="))?.slice("--case=".length);
|
|
const assistantMode = process.argv.includes("--assistant");
|
|
const configuredLoops = Number(process.env.CUSTOM_WIDGET_AI_MAX_LOOPS ?? 10);
|
|
const maxLoops = Number.isInteger(configuredLoops) && configuredLoops > 0 ? Math.min(configuredLoops, 10) : 10;
|
|
let selectableCases = CUSTOM_WIDGET_AI_EVALUATION_CASES.filter((testCase) => !testCase.expectedWidgets?.length);
|
|
if (assistantMode) {
|
|
selectableCases = CUSTOM_WIDGET_AI_EVALUATION_CASES.filter(
|
|
(testCase) =>
|
|
(testCase.sampleResponse !== undefined || testCase.previewResponses?.length) &&
|
|
(testCase.expectations !== undefined || testCase.expectedWidgets?.length),
|
|
);
|
|
}
|
|
let selectedCases = selectableCases;
|
|
if (requestedCase) selectedCases = selectableCases.filter((testCase) => testCase.id === requestedCase);
|
|
if (selectedCases.length === 0) throw new Error(`Unknown AI evaluation case '${requestedCase}'`);
|
|
|
|
const runId = new Date().toISOString().replaceAll(/[:.]/gu, "-");
|
|
const outputRoot = path.resolve(process.cwd(), ".ai-evaluations", runId);
|
|
await mkdir(outputRoot, { recursive: true });
|
|
|
|
function getResultScoreFloor(result: AiEvaluationResult | CustomWidgetAssistantEvaluationResult) {
|
|
if ("judges" in result && result.judges.length > 0) {
|
|
return Math.min(...result.judges.map((judge) => judge.total));
|
|
}
|
|
return result.judge?.total ?? null;
|
|
}
|
|
|
|
const results: Array<AiEvaluationResult | CustomWidgetAssistantEvaluationResult> = [];
|
|
for (const testCase of selectedCases) {
|
|
process.stdout.write(`Evaluating ${testCase.id}...\n`);
|
|
let result: AiEvaluationResult | CustomWidgetAssistantEvaluationResult;
|
|
if (assistantMode) {
|
|
result = await evaluateCustomWidgetAssistantCase({
|
|
testCase,
|
|
apiKey,
|
|
baseUrl: providerBaseUrl,
|
|
outputRoot,
|
|
maxLoops,
|
|
generatorModel,
|
|
judgeModel,
|
|
});
|
|
} else {
|
|
result = await evaluateCustomWidgetCase({
|
|
testCase,
|
|
apiKey,
|
|
baseUrl: providerBaseUrl,
|
|
outputRoot,
|
|
maxLoops,
|
|
generatorModel,
|
|
judgeModel,
|
|
});
|
|
}
|
|
results.push(result);
|
|
if (result.judge) {
|
|
const widgetCount = "widgets" in result ? result.widgets.length : 1;
|
|
const scoreFloor = getResultScoreFloor(result);
|
|
const status = result.judge.verdict === "pass" ? "passed" : "failed";
|
|
process.stdout.write(
|
|
` ${scoreFloor}/100 floor across ${widgetCount} widget(s), ${status} after ${result.attempts} attempt(s)\n`,
|
|
);
|
|
} else {
|
|
process.stdout.write(` failed after ${result.attempts} attempt(s)\n`);
|
|
}
|
|
}
|
|
|
|
const summary = {
|
|
generatedAt: new Date().toISOString(),
|
|
mode: assistantMode ? "assistant-tool-loop" : "manifest",
|
|
providerBaseUrl,
|
|
generatorModel,
|
|
judgeModel,
|
|
results: results.map((result) => ({
|
|
caseId: result.caseId,
|
|
attempts: result.attempts,
|
|
score: getResultScoreFloor(result),
|
|
verdict: result.judge?.verdict ?? "fail",
|
|
categories: result.judge?.categories ?? null,
|
|
errors: result.errors,
|
|
calledTools: "calledTools" in result ? result.calledTools : undefined,
|
|
widgets: "widgets" in result ? result.widgets.length : 1,
|
|
widgetScores: "judges" in result ? result.judges.map((judge) => judge.total) : undefined,
|
|
efficiency: "efficiency" in result ? result.efficiency : undefined,
|
|
outputDirectory: path.relative(process.cwd(), result.outputDirectory),
|
|
})),
|
|
};
|
|
await writeFile(path.join(outputRoot, "summary.json"), JSON.stringify(summary, null, 2), "utf8");
|
|
process.stdout.write(`Results saved to ${path.relative(process.cwd(), outputRoot)}\n`);
|
|
|
|
if (results.some((result) => result.judge?.verdict !== "pass")) process.exitCode = 1;
|