agent-browser
Automate Browser Tasks with AI Agents
متاح أيضًا من: vercel-labs,inferencesh,inf-sh
This skill enables AI agents to automate browser interactions including navigation, form filling, data extraction, and visual recording through a simple @e reference system.
تنزيل ZIP المهارة
رفع في Claude
اذهب إلى Settings → Capabilities → Skills → Upload skill
فعّل وابدأ الاستخدام
اختبرها
استخدام "agent-browser". Open https://example.com and get elements
النتيجة المتوقعة:
Session started with ID: abc123. Elements found: @e1 [a] "Example Domain", @e2 [h1] "Example Domain", @e3 [p] "This domain is for use..."
استخدام "agent-browser". Fill login form and submit
النتيجة المتوقعة:
Filled @e1 with 'user@example.com', filled @e2 with 'password123', clicked @e3. Navigation detected. Re-snapshot recommended.
استخدام "agent-browser". Take screenshot
النتيجة المتوقعة:
Screenshot captured: {base64 encoded image data}, dimensions: 1920x1080
التدقيق الأمني
مخاطر منخفضةThis is a legitimate browser automation skill using Playwright via inference.sh CLI. The static scanner flagged 606 potential issues, but upon evaluation, all findings are false positives: external_commands detections are bash command examples in documentation, network detections are legitimate test URLs, and the heuristic 'dangerous combination' flag describes expected browser automation behavior (network access + code execution + session management). This skill provides standard web automation functionality for AI agents.
مشكلات منخفضة المخاطر (3)
عوامل الخطر
⚙️ الأوامر الخارجية (5)
🌐 الوصول إلى الشبكة (2)
📁 الوصول إلى نظام الملفات (1)
درجة الجودة
ماذا يمكنك بناءه
Automated Form Submission
Fill and submit web forms programmatically for data entry automation
Web Content Extraction
Navigate websites and extract structured data for research or monitoring
Visual Browser Testing
Record browser interactions to create visual documentation or debug issues
جرّب هذه الموجهات
Use the agent-browser skill to open [URL] and get the element references for all interactive elements on the page.
Using session [SESSION_ID], fill the form field with ref @e1 with the text '[TEXT]', then click the button with ref @e2.
Take a full-page screenshot of the current session [SESSION_ID] and save it.
Click element @e1 to navigate to the next page, then take a snapshot and extract all heading text from the new page.
أفضل الممارسات
- Always re-snapshot after navigation to get fresh element references
- Use explicit waits after actions that trigger async loading
- Close sessions when done to free browser resources
- Enable video recording only when needed for debugging
تجنب
- Using element refs from a previous page without re-snapshotting
- Not handling navigation after click actions
- Leaking session IDs or not managing session lifecycle
- Attempting to interact with elements that may not be visible