Modern ad platforms deploy aggressive bot detection to block automated verification tools. hide.ninja gives AI models a fleet of hardware-masked, proxy-isolated browser instances that appear as genuine human visitors — making ad compliance testing on protected sites possible for the first time.
Anti-fraud systems on advertising platforms fingerprint every visitor. Standard headless browsers, Selenium scripts, and Puppeteer sessions are detected and blocked instantly — leaving ad verification teams blind to what real users actually see.
Every browser instance gets a deterministic, unique combination of WebGL/WebGPU renderer strings, Canvas noise, Audio context hashes, font metrics, and screen properties. Anti-bot systems see a genuine device — not an automation tool.
Each browser instance routes through SOCKS5h proxies with full DNS leak protection. Verify that geo-targeted ads render correctly for users in specific countries, cities, and ISPs — all from a single control plane.
Launch hundreds of browser profiles simultaneously, each with its own fingerprint and proxy identity. Run ad verification campaigns across thousands of pages in minutes rather than hours — at a scale that manual QA can never match.
The Ninja MCP server is a remote, hosted service at mcp.hide.ninja that exposes hide.ninja browser instances as tools any MCP-compatible AI model can invoke directly. Authenticate once via OAuth 2.0, and your AI agent can create profiles, assign proxies, launch browsers, and collect results — no local installation, no API keys to manage, just a secure token handshake.
create_profile
Create a new browser profile with a unique hardware fingerprint, proxy, geo-location, and display configuration.
launch_browser
Start a browser instance for a given profile. Returns a session ID for navigation and page interaction.
assign_proxy
Bind a SOCKS5h or HTTP proxy to a profile. Supports geo-targeting by country/city with DNS leak prevention.
capture_page
Navigate to a URL and capture a full-page screenshot or extract DOM content for analysis and verification.
list_profiles
Enumerate all available browser profiles with their fingerprint summary, proxy assignment, and current session status.
close_browser
Gracefully shut down a running browser session. Preserves cookies and session state for subsequent launches.
{ "mcpServers": { "ninja": { "url": "https://mcp.hide.ninja/sse" } } }
No API keys needed in your config — the server uses OAuth 2.0 authorization. Your AI client will open a browser window to mcp.hide.ninja where you sign in with your hide.ninja account. The OAuth token is stored securely by your client.
Point your AI client at mcp.hide.ninja and authorize with OAuth — no packages to install, no keys to copy. Compatible with any MCP-enabled AI client — Claude Desktop, Cursor, Windsurf, or your own agent framework.
Add the Ninja MCP server URL to your AI client's configuration. No npm packages or local installations required — it's a hosted service.
{ "mcpServers": { "ninja": { "url": "https://mcp.hide.ninja/sse" } } }
When your AI client connects for the first time, it will open a browser window to mcp.hide.ninja. Sign in with your hide.ninja account to grant access. The OAuth token is stored securely by your client — you only need to authorize once.
Your AI model now has access to all Ninja MCP tools. Simply describe the ad verification task in natural language.
"Create 5 browser profiles with US residential proxies from different states. Launch each one and visit example-ads.com/campaign/summer-2026. Capture screenshots of the ad placements and report which ads rendered correctly vs. which showed fallback content."
Stop guessing whether your ads render correctly. Verify them with the same browser fingerprints that real users have.
