The vulnerability in apify-mcp-server's fetch-apify-docs tool arises from improper input validation (CWE-20) and incorrect handling of relative path elements (CWE-183). The tool validates allowed documentation domains using String.startsWith() rather than verifying the URL hostname, allowing crafted URLs such as 'https://docs.apify.com.evil.com/' or 'https://[email protected]/' to bypass domain checks. This enables attackers to fetch arbitrary content from unintended hosts and return it to the language model. The vulnerability affects versions prior to 0.9.21 and is addressed by proper hostname validation in version 0.9.21.

An attacker can bypass domain allowlist checks and cause the system to fetch and return arbitrary content from attacker-controlled domains. This can lead to the language model receiving malicious or misleading data, potentially impacting the integrity and reliability of automated data extraction processes. The CVSS 3.1 score is 6.1 (medium), reflecting network attack vector, low complexity, no privileges required, user interaction required, and partial confidentiality and integrity impact.

This vulnerability is fixed in apify-mcp-server version 0.9.21 by correcting the domain validation logic to properly compare URL hostnames. Users should upgrade to version 0.9.21 or later to remediate this issue. No official patch or temporary fix is indicated beyond upgrading. Since this is not a cloud service, remediation depends on user action to update the software.