The Agent Skills Directory

[COMMAND_EXECUTION] (LOW): The MCPConnectionStdio class in scripts/connections.py facilitates local subprocess execution via the Model Context Protocol stdio transport. While this is the primary intended purpose of the skill, it creates a risk of arbitrary command execution if command parameters are influenced by untrusted external data.
[EXTERNAL_DOWNLOADS] (LOW): The MCPConnectionSSE and MCPConnectionHTTP classes in scripts/connections.py enable outbound network connections to arbitrary URLs via Server-Sent Events and HTTP streaming. These are standard MCP capabilities but lack URL whitelisting or sanitization, potentially allowing Server-Side Request Forgery (SSRF).
[DATA_EXFILTRATION] (LOW): The skill provides the necessary primitives (command execution and network access) that could be combined to exfiltrate sensitive local data if an attacker successfully performs an indirect prompt injection against the agent using this library.

mcp-builder