@vercel/agent-eval-playground 0.0.1

package/lib/types.ts

@@ -0,0 +1,113 @@
+/**
+ * Playground types — mirrors the core agent-eval types for the UI.
+ * Kept separate to avoid importing Node.js code into the browser bundle.
+ */
+
+/** Canonical tool names across agents */
+export type ToolName =
+  | "file_read"
+  | "file_write"
+  | "file_edit"
+  | "shell"
+  | "web_fetch"
+  | "web_search"
+  | "glob"
+  | "grep"
+  | "list_dir"
+  | "agent_task"
+  | "unknown";
+
+/** An event in the transcript */
+export interface TranscriptEvent {
+  timestamp?: string;
+  type: "message" | "tool_call" | "tool_result" | "thinking" | "error";
+  role?: "user" | "assistant" | "system";
+  content?: string;
+  tool?: {
+    name: ToolName;
+    originalName: string;
+    args?: Record<string, unknown>;
+    result?: unknown;
+    durationMs?: number;
+    success?: boolean;
+  };
+}
+
+/** Summary statistics derived from the transcript */
+export interface TranscriptSummary {
+  totalTurns: number;
+  toolCalls: Record<ToolName, number>;
+  totalToolCalls: number;
+  webFetches: { url: string; method?: string; status?: number; success?: boolean }[];
+  filesRead: string[];
+  filesModified: string[];
+  shellCommands: { command: string; exitCode?: number; success?: boolean }[];
+  errors: string[];
+  thinkingBlocks: number;
+}
+
+/** A parsed transcript */
+export interface Transcript {
+  agent: string;
+  model?: string;
+  events: TranscriptEvent[];
+  summary: TranscriptSummary;
+  parseSuccess: boolean;
+  parseErrors?: string[];
+}
+
+/** Result of a single eval run */
+export interface EvalRunResult {
+  status: "passed" | "failed";
+  error?: string;
+  duration: number;
+  transcriptPath?: string;
+  transcriptRawPath?: string;
+  outputPaths?: {
+    eval?: string;
+    scripts?: Record<string, string>;
+  };
+  o11y?: TranscriptSummary;
+}
+
+/** Summary of multiple runs for a single eval */
+export interface EvalSummary {
+  name: string;
+  totalRuns: number;
+  passedRuns: number;
+  passRate: number;
+  meanDuration: number;
+}
+
+/** Experiment info returned by the API */
+export interface ExperimentInfo {
+  name: string;
+  timestamps: string[];
+  latestTimestamp: string;
+}
+
+/** Full experiment detail for a specific timestamp */
+export interface ExperimentDetail {
+  startedAt: string;
+  completedAt: string;
+  config: {
+    agent: string;
+    model: string | string[];
+    runs: number;
+    earlyExit: boolean;
+    timeout: number;
+  };
+  evals: EvalSummary[];
+}
+
+/** Run detail with result and optional o11y */
+export interface RunDetail {
+  result: EvalRunResult;
+}
+
+/** Eval fixture info */
+export interface EvalInfo {
+  name: string;
+  prompt: string;
+  files: string[];
+}

package/lib/utils.ts

@@ -0,0 +1,6 @@
+import { clsx, type ClassValue } from "clsx"
+import { twMerge } from "tailwind-merge"
+
+export function cn(...inputs: ClassValue[]) {
+  return twMerge(clsx(inputs))
+}

package/next.config.ts

@@ -0,0 +1,5 @@
+import type { NextConfig } from "next";
+
+const nextConfig: NextConfig = {};
+
+export default nextConfig;

package/package.json

@@ -0,0 +1,51 @@
+{
+  "name": "@vercel/agent-eval-playground",
+  "version": "0.0.1",
+  "description": "Web-based playground for browsing agent-eval experiment results",
+  "bin": {
+    "agent-eval-playground": "./bin.mjs"
+  },
+  "scripts": {
+    "dev": "next dev",
+    "build": "next build",
+    "start": "next start",
+    "release": "changeset publish"
+  },
+  "dependencies": {
+    "@base-ui/react": "^1.1.0",
+    "@remixicon/react": "^4.9.0",
+    "class-variance-authority": "^0.7.1",
+    "clsx": "^2.1.1",
+    "lucide-react": "^0.469.0",
+    "next": "^16.0.0",
+    "radix-ui": "^1.4.3",
+    "react": "^19.2.0",
+    "react-dom": "^19.2.0",
+    "tailwind-merge": "^2.6.0",
+    "tw-animate-css": "^1.4.0"
+  },
+  "devDependencies": {
+    "@tailwindcss/postcss": "^4.0.0",
+    "@types/node": "^22.0.0",
+    "@types/react": "^19.0.0",
+    "@types/react-dom": "^19.0.0",
+    "postcss": "^8.4.49",
+    "tailwindcss": "^4.0.0",
+    "typescript": "^5.6.0"
+  },
+  "files": [
+    "app",
+    "components",
+    "lib",
+    "public",
+    "bin.mjs",
+    "next.config.ts",
+    "tsconfig.json",
+    "postcss.config.mjs",
+    "components.json"
+  ],
+  "publishConfig": {
+    "access": "public"
+  },
+  "license": "MIT"
+}

package/postcss.config.mjs

@@ -0,0 +1,7 @@
+const config = {
+  plugins: {
+    "@tailwindcss/postcss": {},
+  },
+};
+
+export default config;

package/public/vercel.svg

@@ -0,0 +1 @@
+<svg fill="none" xmlns="http://www.w3.org/2000/svg" viewBox="0 0 1155 1000"><path d="m577.3 0 577.4 1000H0z" fill="#fff"/></svg>

package/tsconfig.json

@@ -0,0 +1,42 @@
+{
+  "compilerOptions": {
+    "target": "ES2017",
+    "lib": [
+      "dom",
+      "dom.iterable",
+      "esnext"
+    ],
+    "allowJs": true,
+    "skipLibCheck": true,
+    "strict": true,
+    "noEmit": true,
+    "esModuleInterop": true,
+    "module": "esnext",
+    "moduleResolution": "bundler",
+    "resolveJsonModule": true,
+    "isolatedModules": true,
+    "jsx": "react-jsx",
+    "incremental": true,
+    "plugins": [
+      {
+        "name": "next"
+      }
+    ],
+    "paths": {
+      "@/*": [
+        "./*"
+      ]
+    },
+    "baseUrl": "."
+  },
+  "include": [
+    "next-env.d.ts",
+    "**/*.ts",
+    "**/*.tsx",
+    ".next/types/**/*.ts",
+    ".next/dev/types/**/*.ts"
+  ],
+  "exclude": [
+    "node_modules"
+  ]
+}