The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the terminal tool to execute shell commands, specifically focusing on the codex exec command for autonomous code generation and file system modifications.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data by cloning repositories and checking out pull requests (gh pr checkout) before running Codex analysis or refactoring on that code, which could contain malicious instructions.
Ingestion points: Data enters the environment via git clone, gh pr checkout, and git fetch commands within SKILL.md.
Boundary markers: No boundary markers or specific instructions to ignore embedded commands are present in the task prompts.
Capability inventory: The skill possesses extensive capabilities including file system write access, terminal execution, and network operations through git and gh tools.
Sanitization: There is no evidence of sanitization or content validation before the external code is processed by the Codex agent.
[CREDENTIALS_UNSAFE]: The documentation identifies sensitive file paths where authentication tokens are stored, specifically ~/.hermes/auth.json and ~/.codex/auth.json.
[EXTERNAL_DOWNLOADS]: The skill instructs the installation of the Codex CLI package from OpenAI's official registry, which is a recognized service.

codex