diff --git a/reference/mcp-server.mdx b/reference/mcp-server.mdx
index 9a409e5..93f203a 100644
--- a/reference/mcp-server.mdx
+++ b/reference/mcp-server.mdx
@@ -3,7 +3,7 @@ title: "MCP Server"
 description: "Access Kernel's cloud-based browsers via MCP"
 ---
 
-Our Model Context Protocol (MCP) server lets any compatible AI model or agent launch Chromium browsers, inject context, evaluate JavaScript, and stream DOM snapshots from Kernel's cloud platform — via a single secure endpoint.
+Our Model Context Protocol (MCP) server lets any compatible AI model or agent launch Chromium browsers, execute Playwright code, capture video replays, and automate web interactions from Kernel's cloud platform via a single secure endpoint.
 
 ## Setup Instructions
 
@@ -180,17 +180,21 @@ Configure these values wherever the tool expects MCP server settings.
 ## Tools
 
 ### Browser Automation
+
 - `create_browser` - Launch a new browser session with options (headless, stealth, persistence, timeout, profile)
 - `get_browser` - Get browser session information
 - `list_browsers` - List active browser sessions
 - `delete_browser` - Terminate a browser session
+- `execute_playwright_code` - Execute Playwright/TypeScript code in a fresh browser session with automatic video replay and cleanup
 
 ### Profile Management
+
 - `setup_profile` - Create or update browser profiles with guided setup process
 - `list_profiles` - List all available browser profiles
 - `delete_profile` - Delete browser profile permanently
 
 ### App Management
+
 - `list_apps` - List apps in your Kernel organization with optional filtering
 - `invoke_action` - Execute actions in Kernel apps
 - `get_deployment` - Get deployment status and logs
@@ -198,6 +202,7 @@ Configure these values wherever the tool expects MCP server settings.
 - `get_invocation` - Get action invocation details
 
 ### Documentation & Search
+
 - `search_docs` - Search Kernel platform documentation and guides
 
 ## Resources
@@ -235,6 +240,15 @@ Assistant: I'll create a persistent, stealth-enabled browser that maintains stat
 [Uses create_browser tool with persistence and stealth options]
 ```
 
+### Execute Playwright code dynamically
+
+```
+Human: Go to example.com and get me the page title
+Assistant: I'll execute Playwright code to navigate to the site and retrieve the title.
+[Uses execute_playwright_code tool with code: "await page.goto('https://example.com'); return await page.title();"]
+Returns: { success: true, result: "Example Domain", replay_url: "https://..." }
+```
+
 ### Set up browser profiles for authentication
 
 ```