@testdriverai/agent 7.8.0-canary.15 → 7.8.0-canary.18

package/docs/_scripts/extract-example-urls.js

@@ -1,39 +1,33 @@
 #!/usr/bin/env node
 
 /**
- * Extract Example URLs from CI Logs
- * 
- * Parses vitest output to extract TESTDRIVER_EXAMPLE_URL lines
- * and updates the examples-manifest.json file.
- * 
+ * Extract Example URLs from Test Result JSON Files
+ *
+ * Reads per-test-case JSON result files written by the vitest plugin
+ * to .testdriver/results/ and updates examples-manifest.json.
+ *
  * Usage:
- *   cat ci-log.txt | node extract-example-urls.js
- *   node extract-example-urls.js < ci-log.txt
- *   node extract-example-urls.js --file=ci-log.txt
+ *   node extract-example-urls.js --results-dir=.testdriver/results
  */
 
 const fs = require("fs");
 const path = require("path");
-const readline = require("readline");
 
 const MANIFEST_PATH = path.join(__dirname, "../_data/examples-manifest.json");
 
-// Regex to match TESTDRIVER_EXAMPLE_URL::filename::url (handles optional timestamp prefix from CI logs)
-const URL_PATTERN = /TESTDRIVER_EXAMPLE_URL::([^:]+)::(.+)$/;
-
 // Parse command line arguments
 function parseArgs() {
   const args = process.argv.slice(2);
   const options = {
-    file: null,
+    resultsDir: null,
     help: false,
   };
 
   for (const arg of args) {
     if (arg === "--help" || arg === "-h") {
       options.help = true;
-    } else if (arg.startsWith("--file=")) {
-      options.file = arg.slice(7);
+    } else if (arg.startsWith("--results-dir=")) {
+      options.resultsDir = arg.slice(14);
     }
   }
 
@@ -62,40 +56,51 @@ function saveManifest(manifest) {
   fs.writeFileSync(MANIFEST_PATH, JSON.stringify(manifest, null, 2) + "\n", "utf-8");
 }
 
-// Process a single line and extract URL if present
-function processLine(line, manifest, stats) {
-  const match = line.match(URL_PATTERN);
-  if (match) {
-    const [, filename, url] = match;
-    const isNew = !manifest.examples[filename];
-    
-    manifest.examples[filename] = {
-      url: url.trim(),
-      lastUpdated: new Date().toISOString(),
-    };
-    
-    if (isNew) {
-      stats.added++;
-    } else {
-      stats.updated++;
-    }
-    
-    console.log(`${isNew ? "➕" : "🔄"} ${filename}: ${url}`);
-  }
-}
-
-// Process input stream
-async function processInput(inputStream) {
+// Process JSON result files from .testdriver/results/
+function processResultsDir(resultsDir) {
   const manifest = loadManifest();
   const stats = { added: 0, updated: 0 };
 
-  const rl = readline.createInterface({
-    input: inputStream,
-    crlfDelay: Infinity,
-  });
+  // Look for JSON files under examples/ subdirectories
+  const examplesDir = path.join(resultsDir, "examples");
+  if (!fs.existsSync(examplesDir)) {
+    console.log(`\n⚠️  No examples results found in ${examplesDir}`);
+    return stats;
+  }
 
-  for await (const line of rl) {
-    processLine(line, manifest, stats);
+  // Walk example test directories (e.g., examples/assert.test.mjs/)
+  const testDirs = fs.readdirSync(examplesDir, { withFileTypes: true });
+  for (const entry of testDirs) {
+    if (!entry.isDirectory()) continue;
+    const testDir = path.join(examplesDir, entry.name);
+    const jsonFiles = fs.readdirSync(testDir).filter(f => f.endsWith(".json"));
+
+    for (const jsonFile of jsonFiles) {
+      try {
+        const content = fs.readFileSync(path.join(testDir, jsonFile), "utf-8");
+        const result = JSON.parse(content);
+        const testFileName = path.basename(result.test?.file || result.testFile || entry.name);
+        const url = result.urls?.testRun || result.testRunLink;
+
+        if (!url) continue;
+
+        const isNew = !manifest.examples[testFileName];
+        manifest.examples[testFileName] = {
+          url: url,
+          lastUpdated: result.date || new Date().toISOString(),
+        };
+
+        if (isNew) {
+          stats.added++;
+        } else {
+          stats.updated++;
+        }
+
+        console.log(`${isNew ? "➕" : "🔄"} ${testFileName}: ${url}`);
+      } catch (err) {
+        console.warn(`⚠️  Failed to read ${jsonFile}: ${err.message}`);
+      }
+    }
   }
 
   if (stats.added > 0 || stats.updated > 0) {
@@ -103,7 +108,7 @@ async function processInput(inputStream) {
     console.log(`\n✨ Manifest updated: ${stats.added} added, ${stats.updated} updated`);
     console.log(`📂 Written to: ${MANIFEST_PATH}`);
   } else {
-    console.log("\n⚠️  No TESTDRIVER_EXAMPLE_URL entries found in input");
+    console.log("\n⚠️  No example URLs found in result files");
   }
 
   return stats;
@@ -112,28 +117,24 @@ async function processInput(inputStream) {
 // Show help
 function showHelp() {
   console.log(`
-Extract Example URLs from CI Logs
+Extract Example URLs from Test Result JSON Files
 
 Usage:
-  cat ci-log.txt | node extract-example-urls.js
-  node extract-example-urls.js < ci-log.txt
-  node extract-example-urls.js --file=ci-log.txt
+  node extract-example-urls.js --results-dir=.testdriver/results
 
 Options:
-  --file=<path>   Read from file instead of stdin
-  --help, -h      Show this help message
+  --results-dir=<path>  Path to .testdriver/results directory (required)
+  --help, -h            Show this help message
 
 Description:
-  Parses CI log output looking for lines matching:
-    TESTDRIVER_EXAMPLE_URL::<filename>::<url>
-  
-  Updates docs/_data/examples-manifest.json with the extracted URLs.
+  Reads per-test-case JSON result files from .testdriver/results/examples/
+  and updates docs/_data/examples-manifest.json with the extracted URLs.
   Existing entries are updated, new entries are added.
 `);
 }
 
 // Main function
-async function main() {
+function main() {
   const options = parseArgs();
 
   if (options.help) {
@@ -141,25 +142,19 @@ async function main() {
     process.exit(0);
   }
 
-  console.log("🔍 Extracting example URLs from input...\n");
-
-  let inputStream;
-  if (options.file) {
-    if (!fs.existsSync(options.file)) {
-      console.error(`❌ File not found: ${options.file}`);
-      process.exit(1);
-    }
-    inputStream = fs.createReadStream(options.file);
-  } else {
-    inputStream = process.stdin;
+  if (!options.resultsDir) {
+    console.error("❌ --results-dir is required. Example: --results-dir=.testdriver/results");
+    process.exit(1);
   }
 
-  try {
-    await processInput(inputStream);
-  } catch (error) {
-    console.error(`❌ Error: ${error.message}`);
+  console.log("🔍 Reading example URLs from JSON result files...\n");
+
+  if (!fs.existsSync(options.resultsDir)) {
+    console.error(`❌ Results directory not found: ${options.resultsDir}`);
     process.exit(1);
   }
+
+  processResultsDir(options.resultsDir);
 }
 
 main();

package/docs/docs.json

@@ -87,7 +87,8 @@
             "pages": [
               "/v7/running-tests",
               "/v7/caching",
-              "/v7/ci-cd"
+              "/v7/ci-cd",
+              "/v7/test-results-json"
             ]
           },
           {

package/docs/v7/find.mdx

@@ -50,8 +50,8 @@ const element = await testdriver.find(description, options)
       - `"any"` — No wrapping, uses the description as-is (default behavior)
     </ParamField>
     
-    <ParamField path="zoom" type="boolean" default={false}>
-      Enable two-phase zoom mode for better precision in crowded UIs with many similar elements.
+    <ParamField path="zoom" type="boolean" default={true}>
+      Two-phase zoom mode for better precision in crowded UIs with many similar elements. Enabled by default.
     </ParamField>
     
     <ParamField path="ai" type="object">
@@ -333,14 +333,19 @@ The `timeout` option:
 - Returns the element (check `element.found()` if not throwing on failure)
 - Set to `0` to disable polling and make a single attempt
 
-## Zoom Mode for Crowded UIs
+## Zoom Mode
 
-When dealing with many similar icons or elements clustered together (like browser toolbars), enable `zoom` mode for better precision:
+Zoom mode is **enabled by default**. It uses a two-phase approach for better precision when locating elements, especially in crowded UIs with many similar elements.
+
+To disable zoom for a specific find call, pass `zoom: false`:
 
 ```javascript
-// Enable zoom for better precision in crowded UIs
-const extensionsBtn = await testdriver.find('extensions puzzle icon in Chrome toolbar', { zoom: true });
+// Zoom is on by default — no option needed
+const extensionsBtn = await testdriver.find('extensions puzzle icon in Chrome toolbar');
 await extensionsBtn.click();
+
+// Disable zoom for a specific call if needed
+const largeButton = await testdriver.find('big hero button', { zoom: false });
 ```
 
 ### How Zoom Mode Works
@@ -353,22 +358,11 @@ await extensionsBtn.click();
 This two-phase approach gives the AI a higher-resolution view of the target area, improving accuracy when multiple similar elements are close together.
 
 <Tip>
-  Use `zoom: true` when:
-  - Clicking small icons in toolbars
-  - Selecting from a grid of similar items
-  - Targeting elements in dense UI areas
-  - The default locate is clicking the wrong similar element
-  - You get an AI verification rejection like "The crosshair is located in the empty space of the browser's tab bar/title bar area" — this means the initial locate was imprecise and zoom will help the AI pinpoint the correct element
+  You may want to disable zoom with `zoom: false` when:
+  - Targeting large, isolated elements where the extra precision isn't needed
+  - You want to speed up find calls in simple UIs
 </Tip>
 
-```javascript
-// Without zoom - may click wrong icon in toolbar
-const icon = await testdriver.find('settings icon');
-
-// With zoom - better precision for crowded areas
-const icon = await testdriver.find('settings icon', { zoom: true });
-```
-
 ## Cache Options
 
 Control caching behavior to optimize performance, especially when using dynamic variables in prompts.

package/docs/v7/test-results-json.mdx

@@ -0,0 +1,258 @@
+---
+title: "Test Result JSON"
+sidebarTitle: "Test Result JSON"
+description: "Per-test JSON result files with metadata, versions, and infrastructure details"
+icon: "file-code"
+---
+
+## Overview
+
+TestDriver automatically writes a JSON result file for each test case after it finishes. These files contain comprehensive metadata about the test run, including SDK and runner versions, infrastructure details, interaction statistics, and links to recordings.
+
+Result files are written to:
+
+```
+.testdriver/results/<testFile>/<testName>.json
+```
+
+For example, a test file `tests/login.test.mjs` with a test named `"should log in"` produces:
+
+```
+.testdriver/results/tests/login.test.mjs/should_log_in.json
+```
+
+<Note>
+  Test names are sanitized for filesystem use — special characters are replaced with underscores and names are truncated to 200 characters.
+</Note>
+
+## Enabling
+
+No configuration is required. The JSON files are written automatically by the TestDriver Vitest reporter plugin whenever tests run.
+
+## JSON Schema
+
+Each result file is organized into logical groups:
+
+### `versions`
+
+| Field | Type | Description |
+|---|---|---|
+| `versions.sdk` | `string \| null` | TestDriver SDK version (e.g. `"7.8.0"`) |
+| `versions.vitest` | `string \| null` | Vitest version used to run the test |
+| `versions.api` | `string \| null` | TestDriver API server version |
+| `versions.runnerBefore` | `string \| null` | Runner version at sandbox start |
+| `versions.runnerAfter` | `string \| null` | Runner version after auto-update |
+| `versions.runnerWasUpdated` | `boolean` | Whether the runner was auto-updated during provisioning |
+
+### `test`
+
+| Field | Type | Description |
+|---|---|---|
+| `test.file` | `string \| null` | Relative path to the test file |
+| `test.name` | `string \| null` | Name of the test case |
+| `test.suite` | `string \| null` | Name of the parent `describe` block |
+| `test.passed` | `boolean` | Whether the test passed |
+| `test.caseId` | `string \| null` | Database ID for this test case |
+| `test.runId` | `string \| null` | Database ID for the overall test run |
+| `test.error` | `string \| null` | Error message if the test failed |
+| `test.errorStack` | `string \| null` | Error stack trace if the test failed |
+
+### `urls`
+
+| Field | Type | Description |
+|---|---|---|
+| `urls.api` | `string \| null` | API root URL used for this test |
+| `urls.console` | `string \| null` | TestDriver console base URL |
+| `urls.vnc` | `string \| null` | VNC URL for the sandbox |
+| `urls.testRun` | `string \| null` | Direct link to this test case in the console |
+
+### `replay`
+
+The `replay` object contains the recording replay URL and derived embed links. The `gifUrl` and `embedUrl` are generated automatically from the replay URL.
+
+| Field | Type | Description |
+|---|---|---|
+| `replay.url` | `string \| null` | Recording replay URL |
+| `replay.gifUrl` | `string \| null` | Animated GIF thumbnail of the recording |
+| `replay.embedUrl` | `string \| null` | Embeddable replay URL (appends `&embed=true`) |
+| `replay.markdown` | `string \| null` | Ready-to-use Markdown embed with GIF linking to the replay |
+
+The `replay.markdown` field produces a clickable GIF badge you can paste directly into PR comments, README files, or issue descriptions:
+
+```markdown
+[![Test Recording](https://api.testdriver.ai/replay/abc123/gif?shareKey=xyz)](https://console.testdriver.ai/replay/abc123?share=xyz)
+```
+
+### `date`
+
+| Field | Type | Description |
+|---|---|---|
+| `date` | `string` | ISO 8601 timestamp when the test finished |
+
+### `team`
+
+| Field | Type | Description |
+|---|---|---|
+| `team.id` | `string \| null` | Team ID from the sandbox |
+| `team.sessionId` | `string \| null` | SDK session ID |
+
+### `infrastructure`
+
+| Field | Type | Description |
+|---|---|---|
+| `infrastructure.sandboxId` | `string \| null` | Sandbox instance ID |
+| `infrastructure.instanceId` | `string \| null` | Instance ID |
+| `infrastructure.os` | `string \| null` | Operating system of the sandbox (`"linux"` or `"windows"`) |
+| `infrastructure.amiId` | `string \| null` | AWS AMI ID used for provisioning |
+| `infrastructure.e2bTemplateId` | `string \| null` | E2B template ID used for provisioning |
+| `infrastructure.imageVersion` | `string \| null` | Sandbox image version |
+
+### `realtime`
+
+| Field | Type | Description |
+|---|---|---|
+| `realtime.channel` | `string \| null` | Ably channel name used for communication |
+| `realtime.messageCount` | `number` | Number of messages published to the realtime channel |
+
+### `interactions`
+
+| Field | Type | Description |
+|---|---|---|
+| `interactions.total` | `number` | Total number of interactions recorded |
+| `interactions.cached` | `number` | Number of interactions served from cache |
+| `interactions.byType` | `object` | Breakdown of interactions by type (e.g. `find`, `click`, `assert`) |
+
+## Example Output
+
+```json
+{
+  "sdkVersion": "7.8.0",
+  "vitestVersion": "4.0.0",
+  "apiVersion": "1.45.0",
+  "runnerVersionBefore": "2.1.0",
+  "runnerVersionAfter": "2.1.1",
+  "wasUpdated": true,
+  "apiUrl": "https://api.testdriver.ai",
+  "consoleUrl": "https://console.testdriver.ai",
+  "testRunLink": "https://console.testdriver.ai/runs/abc123/def456",
+  "dashcamUrl": "https://app.dashcam.io/replay/abc123",
+  "vncUrl": "wss://sandbox-123.testdriver.ai/vnc",
+  "date": "2025-01-15T14:30:00.000Z",
+  "team": {
+    "id": "team_abc123",
+    "sessionId": "sess_xyz789"
+  },
+  "infrastructure": {
+    "sandboxId": "sandbox-123",
+    "instanceId": "i-abc123",
+    "os": "linux",
+    "amiId": "ami-0abc123",
+    "e2bTemplateId": null,
+    "imageVersion": "v2.1.0"
+  },
+  "realtime": {
+    "channel": "sandbox:sandbox-123",
+    "messageCount": 42
+  },
+  "interactions": {
+    "total": 15,
+    "cached": 3,
+    "byType": {
+      "find": 8,
+      "click": 5,
+      "assert": 2
+    }
+  }
+}
+```
+
+## Using Result Files in CI
+
+Result files are useful for extracting test metadata in CI pipelines without parsing log output.
+
+### GitHub Actions Example
+
+Use `fromJSON` to parse a result file into a GitHub Actions expression you can reference in subsequent steps:
+
+```yaml
+- name: Run tests
+  run: npx vitest run tests/login.test.mjs
+
+- name: Parse result
+  id: result
+  run: |
+    # Read the first JSON result file
+    FILE=$(find .testdriver/results -name '*.json' | head -n 1)
+    echo "json=$(cat "$FILE")" >> "$GITHUB_OUTPUT"
+
+- name: Comment on PR
+  if: fromJSON(steps.result.outputs.json).test.passed == false
+  uses: actions/github-script@v7
+  with:
+    script: |
+      const result = ${{ steps.result.outputs.json }};
+      await github.rest.issues.createComment({
+        owner: context.repo.owner,
+        repo: context.repo.repo,
+        issue_number: context.issue.number,
+        body: [
+          `❌ **${result.test.name}** failed`,
+          ``,
+          `Error: ${result.test.error}`,
+          ``,
+          result.replay.markdown,
+          ``,
+          `[View full recording](${result.urls.testRun})`
+        ].join('\n')
+      });
+```
+
+You can also load all results into a matrix or iterate over them:
+
+```yaml
+- name: Run tests
+  run: npx vitest run tests/*.test.mjs
+
+- name: Collect results
+  id: results
+  run: |
+    # Merge all result files into a JSON array
+    echo "json=$(find .testdriver/results -name '*.json' -exec cat {} + | jq -s '.')" >> "$GITHUB_OUTPUT"
+
+- name: Summary
+  run: |
+    echo '## Test Results' >> $GITHUB_STEP_SUMMARY
+    RESULTS='${{ steps.results.outputs.json }}'
+    echo "$RESULTS" | jq -r '.[] | "| \(.test.name) | \(if .test.passed then "✅" else "❌" end) | \(.urls.testRun) |"' >> $GITHUB_STEP_SUMMARY
+```
+
+### Reading Results Programmatically
+
+```javascript
+import fs from "fs";
+import path from "path";
+
+const resultsDir = ".testdriver/results";
+
+function readResults(dir) {
+  const results = [];
+  for (const testDir of fs.readdirSync(dir, { recursive: true })) {
+    const fullPath = path.join(dir, testDir);
+    if (fullPath.endsWith(".json") && fs.statSync(fullPath).isFile()) {
+      results.push(JSON.parse(fs.readFileSync(fullPath, "utf-8")));
+    }
+  }
+  return results;
+}
+
+const results = readResults(resultsDir);
+const passed = results.filter(r => r.test.passed);
+const failed = results.filter(r => !r.test.passed);
+
+console.log(`${passed.length} passed, ${failed.length} failed`);
+for (const r of failed) {
+  console.log(`  FAIL: ${r.test.name} — ${r.test.error}`);
+  console.log(`  Recording: ${r.urls.testRun}`);
+  console.log(`  Embed: ${r.replay.markdown}`);
+}
+```

package/examples/scroll-keyboard.test.mjs

@@ -16,7 +16,7 @@ describe("Scroll Keyboard Test", () => {
     // Navigate to https://www.webhamster.com/
     await testdriver.focusApplication("Google Chrome");
     const urlBar = await testdriver.find(
-      "testdriver-sandbox.vercel.app/login, the URL in the omnibox showing the current page", {zoom: true}
+      "the URL in the omnibox", {zoom: true}
     );
     await urlBar.click();
     await testdriver.pressKeys(["ctrl", "a"]);

package/examples/scroll.test.mjs

@@ -12,19 +12,8 @@ import { getDefaults } from "./config.mjs";
 describe("Scroll Test", () => {
   it("should navigate and scroll down the page", async (context) => {
     const testdriver = TestDriver(context, { ...getDefaults(context), headless: true });
-    await testdriver.provision.chrome({ url: 'http://testdriver-sandbox.vercel.app/login' });
+    await testdriver.provision.chrome({ url: 'https://www.webhamster.com/' });
 
-    // Give Chrome a moment to fully render the UI
-    await new Promise(resolve => setTimeout(resolve, 2000));
-
-    // Navigate to webhamster.com - just look for the domain, not the full path
-    const urlBar = await testdriver.find(
-      "testdriver-sandbox.vercel.app, the URL in the address bar",
-    );
-    await urlBar.click();
-    await testdriver.pressKeys(["ctrl", "a"]);
-    await testdriver.type("https://www.webhamster.com/");
-    await testdriver.pressKeys(["enter"]);
 
     // Wait for page to load and click heading
     const heading = await testdriver.find(