The Agent Skills Directory

[COMMAND_EXECUTION]: The SKILL.md file contains instructions for the agent to execute shell commands (codex mcp add) to modify the environment by adding a new Model Context Protocol (MCP) server.
[COMMAND_EXECUTION]: The skill explicitly instructs the agent to attempt privilege escalation. If the initial installation command fails due to permissions or sandboxing, the agent is told to "immediately retry the same command with escalated permissions" and to provide a justification to the system for approval before involving the user.
[EXTERNAL_DOWNLOADS]: The skill attempts to download and install an MCP server from https://developers.openai.com/mcp. While the destination belongs to a well-known service, the automated installation logic combined with instructions for permission escalation poses a risk to system security and integrity.

openai-docs