The Agent Skills Directory

COMMAND_EXECUTION (MEDIUM): The MCPConnectionStdio class in scripts/connections.py enables the execution of local system commands using the stdio_client transport. This is a core feature of the Model Context Protocol (MCP); however, it creates a risk where an attacker could use prompt injection to trick the agent into running unauthorized or harmful commands on the host machine. The severity is reduced to MEDIUM as this is the primary intended function of the script.
EXTERNAL_DOWNLOADS (LOW): The MCPConnectionSSE and MCPConnectionHTTP classes in scripts/connections.py allow the agent to establish outbound network connections to arbitrary URLs for data streaming. Without a strict whitelist of allowed domains, this capability could be used for unauthorized data transmission or interacting with malicious external endpoints.

mcp-builder