The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill takes user-provided questions and incorporates them directly into shell commands without specifying sanitization or escaping mechanisms.
Ingestion points: User questions provided during skill invocation in SKILL.md.
Boundary markers: None; the input is wrapped in double quotes in a shell command, which is insufficient to prevent command injection (e.g., using backticks or subshells).
Capability inventory: Shell command execution via the codex exec tool.
Sanitization: No instructions are provided to sanitize or validate the content of the user's question before execution.
[COMMAND_EXECUTION]: The skill's primary function involves executing a command-line tool (codex) with arguments derived from potentially untrusted user input.
[EXTERNAL_DOWNLOADS]: The documentation encourages the installation of the @openai/codex package via NPM. While the @openai scope is generally associated with a trusted entity, this specific package is not a standard official release and users should verify its authenticity before installation.

codex-qa