0.4.2

* sync version

* 0.4.1

* add release script

README.md

@@ -436,15 +436,15 @@ Core workflow:
 For Claude Code, a [skill](https://platform.claude.com/docs/en/agents-and-tools/agent-skills/best-practices) provides richer context:
 
 ```bash
-cp -r node_modules/agent-browser/skills/browsing-web .claude/skills/
+cp -r node_modules/agent-browser/skills/agent-browser .claude/skills/
 ```
 
 Or download:
 
 ```bash
-mkdir -p .claude/skills/browsing-web
-curl -o .claude/skills/browsing-web/SKILL.md \
-  https://raw.githubusercontent.com/vercel-labs/agent-browser/main/skills/browsing-web/SKILL.md
+mkdir -p .claude/skills/agent-browser
+curl -o .claude/skills/agent-browser/SKILL.md \
+  https://raw.githubusercontent.com/vercel-labs/agent-browser/main/skills/agent-browser/SKILL.md
 ```
 
 ## License

cli/Cargo.lock

@@ -4,7 +4,7 @@ version = 4
 
 [[package]]
 name = "agent-browser"
-version = "0.4.0"
+version = "0.4.2"
 dependencies = [
  "libc",
  "serde",

cli/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "agent-browser"
-version = "0.4.0"
+version = "0.4.2"
 edition = "2021"
 description = "Fast browser automation CLI for AI agents"
 license = "Apache-2.0"

package.json

@@ -1,6 +1,6 @@
 {
   "name": "agent-browser",
-  "version": "0.4.0",
+  "version": "0.4.2",
   "description": "Headless browser automation CLI for AI agents",
   "type": "module",
   "main": "dist/daemon.js",
@@ -15,13 +15,16 @@
   },
   "scripts": {
     "prepare": "husky",
+    "version:sync": "node scripts/sync-version.js",
+    "version": "npm run version:sync && git add cli/Cargo.toml",
     "build": "tsc",
-    "build:native": "cargo build --release --manifest-path cli/Cargo.toml && node scripts/copy-native.js",
-    "build:linux": "docker compose -f docker/docker-compose.yml run --rm build-linux",
-    "build:macos": "(cargo build --release --manifest-path cli/Cargo.toml --target aarch64-apple-darwin & cargo build --release --manifest-path cli/Cargo.toml --target x86_64-apple-darwin & wait) && cp cli/target/aarch64-apple-darwin/release/agent-browser bin/agent-browser-darwin-arm64 && cp cli/target/x86_64-apple-darwin/release/agent-browser bin/agent-browser-darwin-x64",
-    "build:windows": "docker compose -f docker/docker-compose.yml run --rm build-windows",
-    "build:all-platforms": "(npm run build:linux & npm run build:windows & wait) && npm run build:macos",
+    "build:native": "npm run version:sync && cargo build --release --manifest-path cli/Cargo.toml && node scripts/copy-native.js",
+    "build:linux": "npm run version:sync && docker compose -f docker/docker-compose.yml run --rm build-linux",
+    "build:macos": "npm run version:sync && (cargo build --release --manifest-path cli/Cargo.toml --target aarch64-apple-darwin & cargo build --release --manifest-path cli/Cargo.toml --target x86_64-apple-darwin & wait) && cp cli/target/aarch64-apple-darwin/release/agent-browser bin/agent-browser-darwin-arm64 && cp cli/target/x86_64-apple-darwin/release/agent-browser bin/agent-browser-darwin-x64",
+    "build:windows": "npm run version:sync && docker compose -f docker/docker-compose.yml run --rm build-windows",
+    "build:all-platforms": "npm run version:sync && (npm run build:linux & npm run build:windows & wait) && npm run build:macos",
     "build:docker": "docker build -t agent-browser-builder -f docker/Dockerfile.build .",
+    "release": "npm run version:sync && npm run build && npm run build:all-platforms && npm publish",
     "start": "node dist/daemon.js",
     "dev": "tsx src/daemon.ts",
     "typecheck": "tsc --noEmit",

scripts/sync-version.js

@@ -0,0 +1,43 @@
+#!/usr/bin/env node
+
+/**
+ * Syncs the version from package.json to all other config files.
+ * Run this script before building or releasing.
+ */
+
+import { readFileSync, writeFileSync } from "fs";
+import { dirname, join } from "path";
+import { fileURLToPath } from "url";
+
+const __dirname = dirname(fileURLToPath(import.meta.url));
+const rootDir = join(__dirname, "..");
+
+// Read version from package.json (single source of truth)
+const packageJson = JSON.parse(
+  readFileSync(join(rootDir, "package.json"), "utf-8")
+);
+const version = packageJson.version;
+
+console.log(`Syncing version ${version} to all config files...`);
+
+// Update Cargo.toml
+const cargoTomlPath = join(rootDir, "cli/Cargo.toml");
+let cargoToml = readFileSync(cargoTomlPath, "utf-8");
+const cargoVersionRegex = /^version\s*=\s*"[^"]*"/m;
+const newCargoVersion = `version = "${version}"`;
+
+if (cargoVersionRegex.test(cargoToml)) {
+  const oldMatch = cargoToml.match(cargoVersionRegex)?.[0];
+  if (oldMatch !== newCargoVersion) {
+    cargoToml = cargoToml.replace(cargoVersionRegex, newCargoVersion);
+    writeFileSync(cargoTomlPath, cargoToml);
+    console.log(`  Updated cli/Cargo.toml: ${oldMatch} -> ${newCargoVersion}`);
+  } else {
+    console.log(`  cli/Cargo.toml already up to date`);
+  }
+} else {
+  console.error("  Could not find version field in cli/Cargo.toml");
+  process.exit(1);
+}
+
+console.log("Version sync complete.");

skills/agent-browser/SKILL.md

@@ -1,5 +1,5 @@
 ---
-name: browsing-web
+name: agent-browser
 description: Automates browser interactions for web testing, form filling, screenshots, and data extraction. Use when the user needs to navigate websites, interact with web pages, fill forms, take screenshots, test web applications, or extract information from web pages.
 ---
 

src/actions.ts

@@ -111,6 +111,47 @@ interface SnapshotData {
   refs?: Record<string, { role: string; name?: string }>;
 }
 
+/**
+ * Convert Playwright errors to AI-friendly messages
+ */
+function toAIFriendlyError(error: unknown, selector: string): Error {
+  const message = error instanceof Error ? error.message : String(error);
+
+  // Handle strict mode violation (multiple elements match)
+  if (message.includes('strict mode violation')) {
+    // Extract count if available
+    const countMatch = message.match(/resolved to (\d+) elements/);
+    const count = countMatch ? countMatch[1] : 'multiple';
+
+    return new Error(
+      `Selector "${selector}" matched ${count} elements. ` +
+        `Run 'snapshot' to get updated refs, or use a more specific CSS selector.`
+    );
+  }
+
+  // Handle element not found
+  if (
+    message.includes('waiting for') &&
+    (message.includes('to be visible') || message.includes('Timeout'))
+  ) {
+    return new Error(
+      `Element "${selector}" not found or not visible. ` +
+        `Run 'snapshot' to see current page elements.`
+    );
+  }
+
+  // Handle element not interactable
+  if (message.includes('intercepts pointer events') || message.includes('not visible')) {
+    return new Error(
+      `Element "${selector}" is not interactable (may be hidden or covered). ` +
+        `Try scrolling it into view or check if a modal/overlay is blocking it.`
+    );
+  }
+
+  // Return original error for unknown cases
+  return error instanceof Error ? error : new Error(message);
+}
+
 /**
  * Execute a command and return a response
  */
@@ -384,11 +425,15 @@ async function handleClick(command: ClickCommand, browser: BrowserManager): Prom
   // Support both refs (@e1) and regular selectors
   const locator = browser.getLocator(command.selector);
 
-  await locator.click({
-    button: command.button,
-    clickCount: command.clickCount,
-    delay: command.delay,
-  });
+  try {
+    await locator.click({
+      button: command.button,
+      clickCount: command.clickCount,
+      delay: command.delay,
+    });
+  } catch (error) {
+    throw toAIFriendlyError(error, command.selector);
+  }
 
   return successResponse(command.id, { clicked: true });
 }
@@ -396,13 +441,17 @@ async function handleClick(command: ClickCommand, browser: BrowserManager): Prom
 async function handleType(command: TypeCommand, browser: BrowserManager): Promise<Response> {
   const locator = browser.getLocator(command.selector);
 
-  if (command.clear) {
-    await locator.fill('');
-  }
+  try {
+    if (command.clear) {
+      await locator.fill('');
+    }
 
-  await locator.pressSequentially(command.text, {
-    delay: command.delay,
-  });
+    await locator.pressSequentially(command.text, {
+      delay: command.delay,
+    });
+  } catch (error) {
+    throw toAIFriendlyError(error, command.selector);
+  }
 
   return successResponse(command.id, { typed: true });
 }
@@ -556,14 +605,22 @@ async function handleSelect(command: SelectCommand, browser: BrowserManager): Pr
   const locator = browser.getLocator(command.selector);
   const values = Array.isArray(command.values) ? command.values : [command.values];
 
-  await locator.selectOption(values);
+  try {
+    await locator.selectOption(values);
+  } catch (error) {
+    throw toAIFriendlyError(error, command.selector);
+  }
 
   return successResponse(command.id, { selected: values });
 }
 
 async function handleHover(command: HoverCommand, browser: BrowserManager): Promise<Response> {
   const locator = browser.getLocator(command.selector);
-  await locator.hover();
+  try {
+    await locator.hover();
+  } catch (error) {
+    throw toAIFriendlyError(error, command.selector);
+  }
 
   return successResponse(command.id, { hovered: true });
 }
@@ -643,26 +700,42 @@ async function handleWindowNew(
 
 async function handleFill(command: FillCommand, browser: BrowserManager): Promise<Response> {
   const locator = browser.getLocator(command.selector);
-  await locator.fill(command.value);
+  try {
+    await locator.fill(command.value);
+  } catch (error) {
+    throw toAIFriendlyError(error, command.selector);
+  }
   return successResponse(command.id, { filled: true });
 }
 
 async function handleCheck(command: CheckCommand, browser: BrowserManager): Promise<Response> {
   const locator = browser.getLocator(command.selector);
-  await locator.check();
+  try {
+    await locator.check();
+  } catch (error) {
+    throw toAIFriendlyError(error, command.selector);
+  }
   return successResponse(command.id, { checked: true });
 }
 
 async function handleUncheck(command: UncheckCommand, browser: BrowserManager): Promise<Response> {
   const locator = browser.getLocator(command.selector);
-  await locator.uncheck();
+  try {
+    await locator.uncheck();
+  } catch (error) {
+    throw toAIFriendlyError(error, command.selector);
+  }
   return successResponse(command.id, { unchecked: true });
 }
 
 async function handleUpload(command: UploadCommand, browser: BrowserManager): Promise<Response> {
   const locator = browser.getLocator(command.selector);
   const files = Array.isArray(command.files) ? command.files : [command.files];
-  await locator.setInputFiles(files);
+  try {
+    await locator.setInputFiles(files);
+  } catch (error) {
+    throw toAIFriendlyError(error, command.selector);
+  }
   return successResponse(command.id, { uploaded: files });
 }
 
@@ -671,13 +744,21 @@ async function handleDoubleClick(
   browser: BrowserManager
 ): Promise<Response> {
   const locator = browser.getLocator(command.selector);
-  await locator.dblclick();
+  try {
+    await locator.dblclick();
+  } catch (error) {
+    throw toAIFriendlyError(error, command.selector);
+  }
   return successResponse(command.id, { clicked: true });
 }
 
 async function handleFocus(command: FocusCommand, browser: BrowserManager): Promise<Response> {
   const locator = browser.getLocator(command.selector);
-  await locator.focus();
+  try {
+    await locator.focus();
+  } catch (error) {
+    throw toAIFriendlyError(error, command.selector);
+  }
   return successResponse(command.id, { focused: true });
 }
 

src/browser.ts

@@ -95,12 +95,20 @@ export class BrowserManager {
 
     const page = this.getPage();
 
-    // Parse the selector and create locator
+    // Build locator with exact: true to avoid substring matches
+    let locator: Locator;
     if (refData.name) {
-      return page.getByRole(refData.role as any, { name: refData.name });
+      locator = page.getByRole(refData.role as any, { name: refData.name, exact: true });
     } else {
-      return page.getByRole(refData.role as any);
+      locator = page.getByRole(refData.role as any);
     }
+
+    // If an nth index is stored (for disambiguation), use it
+    if (refData.nth !== undefined) {
+      locator = locator.nth(refData.nth);
+    }
+
+    return locator;
   }
 
   /**

src/snapshot.ts

@@ -24,6 +24,8 @@ export interface RefMap {
     selector: string;
     role: string;
     name?: string;
+    /** Index for disambiguation when multiple elements have same role+name */
+    nth?: number;
   };
 }
 
@@ -129,7 +131,7 @@ const STRUCTURAL_ROLES = new Set([
 function buildSelector(role: string, name?: string): string {
   if (name) {
     const escapedName = name.replace(/"/g, '\\"');
-    return `getByRole('${role}', { name: "${escapedName}" })`;
+    return `getByRole('${role}', { name: "${escapedName}", exact: true })`;
   }
   return `getByRole('${role}')`;
 }
@@ -161,12 +163,60 @@ export async function getEnhancedSnapshot(
   return { tree: enhancedTree, refs };
 }
 
+/**
+ * Track role+name combinations to detect duplicates
+ */
+interface RoleNameTracker {
+  counts: Map<string, number>;
+  /** Maps role+name key to array of ref IDs that use it */
+  refsByKey: Map<string, string[]>;
+  getKey(role: string, name?: string): string;
+  getNextIndex(role: string, name?: string): number;
+  trackRef(role: string, name: string | undefined, ref: string): void;
+  /** Get all role+name keys that have duplicates */
+  getDuplicateKeys(): Set<string>;
+}
+
+function createRoleNameTracker(): RoleNameTracker {
+  const counts = new Map<string, number>();
+  const refsByKey = new Map<string, string[]>();
+  return {
+    counts,
+    refsByKey,
+    getKey(role: string, name?: string): string {
+      return `${role}:${name ?? ''}`;
+    },
+    getNextIndex(role: string, name?: string): number {
+      const key = this.getKey(role, name);
+      const current = counts.get(key) ?? 0;
+      counts.set(key, current + 1);
+      return current;
+    },
+    trackRef(role: string, name: string | undefined, ref: string): void {
+      const key = this.getKey(role, name);
+      const refs = refsByKey.get(key) ?? [];
+      refs.push(ref);
+      refsByKey.set(key, refs);
+    },
+    getDuplicateKeys(): Set<string> {
+      const duplicates = new Set<string>();
+      for (const [key, refs] of refsByKey) {
+        if (refs.length > 1) {
+          duplicates.add(key);
+        }
+      }
+      return duplicates;
+    },
+  };
+}
+
 /**
  * Process ARIA snapshot: add refs and apply filters
  */
 function processAriaTree(ariaTree: string, refs: RefMap, options: SnapshotOptions): string {
   const lines = ariaTree.split('\n');
   const result: string[] = [];
+  const tracker = createRoleNameTracker();
 
   // For interactive-only mode, we collect just interactive elements
   if (options.interactive) {
@@ -179,31 +229,43 @@ function processAriaTree(ariaTree: string, refs: RefMap, options: SnapshotOption
 
       if (INTERACTIVE_ROLES.has(roleLower)) {
         const ref = nextRef();
+        const nth = tracker.getNextIndex(roleLower, name);
+        tracker.trackRef(roleLower, name, ref);
         refs[ref] = {
           selector: buildSelector(roleLower, name),
           role: roleLower,
           name,
+          nth, // Always store nth, we'll use it for duplicates
         };
 
         let enhanced = `- ${role}`;
         if (name) enhanced += ` "${name}"`;
         enhanced += ` [ref=${ref}]`;
+        // Only show nth in output if it's > 0 (for readability)
+        if (nth > 0) enhanced += ` [nth=${nth}]`;
         if (suffix && suffix.includes('[')) enhanced += suffix;
 
         result.push(enhanced);
       }
     }
+
+    // Post-process: remove nth from refs that don't have duplicates
+    removeNthFromNonDuplicates(refs, tracker);
+
     return result.join('\n') || '(no interactive elements)';
   }
 
   // Normal processing with depth/compact filters
   for (const line of lines) {
-    const processed = processLine(line, refs, options);
+    const processed = processLine(line, refs, options, tracker);
     if (processed !== null) {
       result.push(processed);
     }
   }
 
+  // Post-process: remove nth from refs that don't have duplicates
+  removeNthFromNonDuplicates(refs, tracker);
+
   // If compact mode, remove empty structural elements
   if (options.compact) {
     return compactTree(result.join('\n'));
@@ -212,6 +274,22 @@ function processAriaTree(ariaTree: string, refs: RefMap, options: SnapshotOption
   return result.join('\n');
 }
 
+/**
+ * Remove nth from refs that ended up not having duplicates
+ * This keeps single-element locators simple (no unnecessary .nth(0))
+ */
+function removeNthFromNonDuplicates(refs: RefMap, tracker: RoleNameTracker): void {
+  const duplicateKeys = tracker.getDuplicateKeys();
+
+  for (const [ref, data] of Object.entries(refs)) {
+    const key = tracker.getKey(data.role, data.name);
+    if (!duplicateKeys.has(key)) {
+      // Not a duplicate, remove nth to keep locator simple
+      delete refs[ref].nth;
+    }
+  }
+}
+
 /**
  * Get indentation level (number of spaces / 2)
  */
@@ -223,7 +301,12 @@ function getIndentLevel(line: string): number {
 /**
  * Process a single line: add ref if needed, filter if requested
  */
-function processLine(line: string, refs: RefMap, options: SnapshotOptions): string | null {
+function processLine(
+  line: string,
+  refs: RefMap,
+  options: SnapshotOptions,
+  tracker: RoleNameTracker
+): string | null {
   const depth = getIndentLevel(line);
 
   // Check max depth
@@ -273,17 +356,22 @@ function processLine(line: string, refs: RefMap, options: SnapshotOptions): stri
 
   if (shouldHaveRef) {
     const ref = nextRef();
+    const nth = tracker.getNextIndex(roleLower, name);
+    tracker.trackRef(roleLower, name, ref);
 
     refs[ref] = {
       selector: buildSelector(roleLower, name),
       role: roleLower,
       name,
+      nth, // Always store nth, we'll clean up non-duplicates later
     };
 
     // Build enhanced line with ref
     let enhanced = `${prefix}${role}`;
     if (name) enhanced += ` "${name}"`;
     enhanced += ` [ref=${ref}]`;
+    // Only show nth in output if it's > 0 (for readability)
+    if (nth > 0) enhanced += ` [nth=${nth}]`;
     if (suffix) enhanced += suffix;
 
     return enhanced;