# URL SSRF Prevention Validator MCP for AI Agents AI Agent Connect > URL SSRF Prevention Validator prevents Server-Side Request Forgery attacks by verifying URL schemes and blocking access to private, internal, or loopback IP ranges. It acts as a security gatekeeper, ensuring your AI agents only interact with safe, public web resources and never accidentally probe your local network or sensitive internal infrastructure. ## Overview - **Category:** security - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_b9lgFhNxKWewQBquXykda4HVnbGMgva0k6pVvTmr/ai-agent-connect - **Tags:** ssrf, security, url-validation, network-safety, ip-filtering ## Description When you give an AI agent the ability to browse the web or fetch data from URLs, you're essentially giving it a key to your network. Without proper guardrails, a malicious link or a clever prompt could trick that agent into scanning your internal servers, accessing private databases, or hitting local services that were never meant to be public. This is a classic Server-Side Request Forgery vulnerability, and it's a massive risk for anyone running automated agents. This MCP acts as a dedicated security layer that sits between your agent and the internet. It intercepts every URL request to check if the target is a legitimate public resource. It uses precise math to identify and block any attempts to reach private IP addresses, loopback addresses, or restricted local ranges. Instead of manually writing complex validation logic for every single tool you build, you can just connect this to your workflow through the Vinkius catalog. It keeps your agent focused on the public web while keeping your internal network completely invisible and protected from unauthorized probing. ## Tools ### check_ip_safety Checks if an IP address belongs to a private or restricted range. This prevents agents from reaching your internal network. ### parse_url_components Breaks a URL down into its individual parts. This allows for a granular inspection of the host, path, and scheme. ### validate_url Runs a full safety check on a URL. It verifies the scheme and IP range to ensure the target is a safe public resource. ## Prompt Examples **Prompt:** ``` Is this URL safe to visit: http://169.254.169.254/latest/meta-data/? ``` **Response:** ``` No, that URL is not safe. It targets a link-local address used for cloud metadata services. Accessing this could expose sensitive cloud credentials or instance information. You should only use public, non-private IP addresses. ``` **Prompt:** ``` Check if I can safely use this link: file:///etc/passwd ``` **Response:** ``` No, that link is not safe. The `file://` scheme is prohibited because it attempts to access local system files. This MCP is designed to restrict access to web-based protocols like HTTP and HTTPS to protect your local environment. ``` **Prompt:** ``` Can my agent access https://vinkius.com? ``` **Response:** ``` Yes, that is a safe, public URL. It uses a standard HTTPS scheme and points to a valid public domain, making it safe for your agent to access. ``` ## Capabilities ### Block internal network probing Prevents agents from hitting private IP addresses or local services. ### Verify URL schemes Ensures agents only use safe protocols like HTTPS and blocks dangerous ones like file or gopher. ### Identify private IP ranges Uses deterministic math to catch attempts to access loopback and private network segments. ### Deconstruct URL components Breaks down web addresses to inspect every part of the request for hidden threats. ### Validate web resource safety Confirms a target is a safe, public destination before any connection is made. ## Use Cases ### Securing autonomous web researchers A researcher agent is tasked with finding data on the web. It uses the validator to ensure it doesn't accidentally try to access a local admin panel on the host machine. ### Protecting cloud microservices An agent fetching data from external APIs is prevented from being tricked into hitting internal metadata services or private cloud endpoints. ### Safe document processing When an agent processes links found in documents, it verifies the URL is a standard web link and not a local file path attempt. ### Preventing malicious prompt injection A user tries to trick an agent into leaking internal data by providing a URL pointing to a local service. The validator catches and blocks the request immediately. ## Benefits - Stop accidental network probing by automatically blocking requests to private IP ranges. - Prevent unauthorized file access by restricting dangerous URL schemes like file://. - Reduce security debt by using a standardized validation method instead of custom regex. - Ensure agent reliability by confirming URLs are valid and public before they attempt to fetch data. - Secure your local environment by preventing loopback address exploitation. ## Frequently Asked Questions **How does the URL SSRF Prevention Validator protect my network?** It acts as a gatekeeper that checks every URL your agent tries to access. It blocks any requests that point to private IP addresses, local loopback addresses, or dangerous file schemes, ensuring your agent stays on the public web. **Can I use the URL SSRF Prevention Validator with Claude Desktop?** Yes, you can connect this MCP to Claude Desktop, Cursor, Windsurf, and any other MCP-compatible client to secure your web-based workflows. **Will this MCP stop an agent from reading my local files?** Yes. By validating the URL scheme, it prevents agents from using protocols like `file://` to access sensitive files on your local machine or server. **Does the URL SSRF Prevention Validator work for all types of web requests?** It is designed to validate the safety of the destination. It checks the IP address and the protocol to ensure the target is a legitimate, public web resource. **What happens if an agent tries to access a blocked internal IP?** The request will be identified as unsafe, and the agent will be notified that the target is a private or restricted address, preventing the connection from ever being made. **How does this tool prevent SSRF attacks?** The `validate_url` tool checks the URL scheme to ensure it is web-safe and uses bitwise integer math to block any requests targeting private or loopback IP ranges. **Which IP ranges are blocked?** It blocks loopback addresses, Class A, B, and C private networks (10.x.x.x, 172.16.x.x, 192.168.x.x), and 0.0.0.0. **Can I use this with Claude Desktop?** Yes, this MCP server is compatible with Claude Desktop and other MCP-compliant clients via Vinkius Edge.