Executive brief
AIAnytime Awesome-MCP-Server is a collection of Model Context Protocol (MCP) servers that allow AI agents to interact with external data sources like Wikipedia. A security flaw in the Wikipedia-related tools allows an attacker to force the server to make unauthorized network requests. This could be used to scan internal networks, access sensitive cloud metadata, or interact with other private services that are not intended to be public.
Technical details
A Server-Side Request Forgery (SSRF) vulnerability exists in the mcp-wiki and wiki-summary components of the Awesome-MCP-Server project. The root cause is found in 'mcp-wiki/src/mcp_wiki/server.py' and 'streamlit as an MCP Host/ollama_server.py', where the 'read_wikipedia_article' and 'summarize_wikipedia_article' tools accept a 'url' parameter. The application only performs a minimal prefix check (url.startswith("http")) before passing the URL to the Python 'requests.get()' function. A remote attacker with the ability to invoke these tools (e.g., via a hijacked MCP client or indirect prompt injection) can force the server to perform GET requests against arbitrary internal or external IP addresses and hostnames. As of the advisory date, the project uses a rolling release model and has not yet responded to the issue reports.
Affected products
- AIAnytime Awesome-MCP-Server up to a884bb51bcd99e08e14fd712c749d55d9d9a13ab
Timeline
- 2026-06-05: disclosed: Issue reported to the project maintainers via GitHub issues #34 and #35
- 2026-07-05: advisory: CVE-2026-14748 published by VulDB