agent-regression-lab 0.1.0

package/dist/ui/server.js

@@ -0,0 +1,147 @@
+import { build } from "esbuild";
+import { createServer } from "node:http";
+import { readFileSync, existsSync } from "node:fs";
+import { extname, resolve } from "node:path";
+import { ensureDir } from "../lib/fs.js";
+import { getRunErrorDetail } from "../runOutput.js";
+import { Storage } from "../storage.js";
+const UI_ROOT = resolve("artifacts", "ui");
+const ASSETS_ROOT = resolve(UI_ROOT, "assets");
+const PORT = 4173;
+export async function startUiServer() {
+    await buildUiAssets();
+    const server = createServer((request, response) => {
+        void handleRequest(request, response);
+    });
+    await new Promise((resolvePromise) => {
+        server.listen(PORT, "127.0.0.1", () => resolvePromise());
+    });
+    console.log(`UI available at http://127.0.0.1:${PORT}`);
+}
+async function handleRequest(request, response) {
+    const url = new URL(request.url ?? "/", `http://${request.headers.host ?? `127.0.0.1:${PORT}`}`);
+    if (url.pathname.startsWith("/api/")) {
+        handleApi(url, response);
+        return;
+    }
+    if (url.pathname.startsWith("/assets/")) {
+        serveStatic(resolve(UI_ROOT, `.${url.pathname}`), response);
+        return;
+    }
+    response.writeHead(200, { "Content-Type": "text/html; charset=utf-8" });
+    response.end(renderHtml());
+}
+function handleApi(url, response) {
+    const storage = new Storage();
+    try {
+        if (url.pathname === "/api/runs") {
+            const payload = storage.listRuns({
+                suite: url.searchParams.get("suite") || undefined,
+                status: url.searchParams.get("status") ?? undefined,
+                provider: url.searchParams.get("provider") || undefined,
+            });
+            sendJson(response, 200, { runs: payload });
+            return;
+        }
+        if (url.pathname.startsWith("/api/runs/")) {
+            const runId = decodeURIComponent(url.pathname.slice("/api/runs/".length));
+            const bundle = storage.getRun(runId);
+            if (!bundle) {
+                sendJson(response, 404, { error: `Run '${runId}' not found.` });
+                return;
+            }
+            sendJson(response, 200, {
+                run: bundle.run,
+                agentVersion: bundle.agentVersion,
+                evaluatorResults: bundle.evaluatorResults,
+                toolCalls: bundle.toolCalls,
+                traceEvents: bundle.traceEvents,
+                errorDetail: getRunErrorDetail(bundle),
+            });
+            return;
+        }
+        if (url.pathname === "/api/compare") {
+            const baseline = url.searchParams.get("baseline");
+            const candidate = url.searchParams.get("candidate");
+            if (!baseline || !candidate) {
+                sendJson(response, 400, { error: "Both 'baseline' and 'candidate' query params are required." });
+                return;
+            }
+            const comparison = storage.compareRuns(baseline, candidate);
+            sendJson(response, 200, {
+                baseline: {
+                    ...comparison.baseline,
+                    errorDetail: getRunErrorDetail(comparison.baseline),
+                },
+                candidate: {
+                    ...comparison.candidate,
+                    errorDetail: getRunErrorDetail(comparison.candidate),
+                },
+                notes: comparison.notes,
+                deltas: comparison.deltas,
+                evaluatorDiffs: comparison.evaluatorDiffs,
+                toolDiffs: comparison.toolDiffs,
+            });
+            return;
+        }
+        sendJson(response, 404, { error: "Not found." });
+    }
+    catch (error) {
+        sendJson(response, 500, { error: error instanceof Error ? error.message : String(error) });
+    }
+}
+async function buildUiAssets() {
+    ensureDir(ASSETS_ROOT);
+    await build({
+        entryPoints: [resolve("src", "ui", "client.tsx")],
+        outdir: ASSETS_ROOT,
+        bundle: true,
+        format: "esm",
+        splitting: false,
+        platform: "browser",
+        sourcemap: false,
+        logLevel: "silent",
+        loader: {
+            ".css": "css",
+        },
+    });
+}
+function serveStatic(path, response) {
+    if (!existsSync(path)) {
+        response.writeHead(404, { "Content-Type": "text/plain; charset=utf-8" });
+        response.end("Not found");
+        return;
+    }
+    const contentType = getContentType(path);
+    response.writeHead(200, { "Content-Type": contentType });
+    response.end(readFileSync(path));
+}
+function sendJson(response, status, payload) {
+    response.writeHead(status, { "Content-Type": "application/json; charset=utf-8" });
+    response.end(JSON.stringify(payload));
+}
+function renderHtml() {
+    return `<!doctype html>
+<html lang="en">
+  <head>
+    <meta charset="utf-8" />
+    <meta name="viewport" content="width=device-width, initial-scale=1" />
+    <title>Agent Regression Lab Alpha</title>
+    <link rel="stylesheet" href="/assets/client.css" />
+  </head>
+  <body>
+    <div id="root"></div>
+    <script type="module" src="/assets/client.js"></script>
+  </body>
+</html>`;
+}
+function getContentType(path) {
+    switch (extname(path)) {
+        case ".js":
+            return "application/javascript; charset=utf-8";
+        case ".css":
+            return "text/css; charset=utf-8";
+        default:
+            return "text/plain; charset=utf-8";
+    }
+}

package/package.json

@@ -0,0 +1,53 @@
+{
+  "name": "agent-regression-lab",
+  "version": "0.1.0",
+  "private": false,
+  "description": "Local-first scenario-based evaluation harness for AI agents.",
+  "license": "MIT",
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/YakshithK/agent-regression-lab.git"
+  },
+  "homepage": "https://github.com/YakshithK/agent-regression-lab#readme",
+  "bugs": {
+    "url": "https://github.com/YakshithK/agent-regression-lab/issues"
+  },
+  "keywords": [
+    "ai",
+    "agent",
+    "evaluation",
+    "cli",
+    "regression"
+  ],
+  "type": "module",
+  "bin": {
+    "agentlab": "./dist/index.js"
+  },
+  "files": [
+    "dist",
+    "README.md"
+  ],
+  "engines": {
+    "node": ">=22"
+  },
+  "scripts": {
+    "build": "tsc -p tsconfig.json",
+    "check": "tsc -p tsconfig.json --noEmit",
+    "test": "tsx --test tests/**/*.test.ts",
+    "smoke:cli": "npm run build && node dist/index.js --help && node dist/index.js version",
+    "start": "tsx src/index.ts",
+    "run": "tsx src/index.ts"
+  },
+  "dependencies": {
+    "@types/node": "^25.5.0",
+    "@types/react": "^19.2.14",
+    "@types/react-dom": "^19.2.3",
+    "esbuild": "^0.27.4",
+    "openai": "^6.33.0",
+    "react": "^19.2.4",
+    "react-dom": "^19.2.4",
+    "tsx": "^4.21.0",
+    "typescript": "^6.0.2",
+    "yaml": "^2.8.3"
+  }
+}