The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute various shell commands including codex review, git, and gh to manage code review targets and fetch branch metadata. It specifically uses sh -c to run user-defined test commands in parallel with the review process.
[PROMPT_INJECTION]: The skill contains an indirect prompt injection surface (Category 8).
Ingestion points: Source code and PR metadata are ingested via codex review and gh pr view (SKILL.md).
Boundary markers: The skill includes instructions to "Treat review output as advisory" and "Verify every finding by reading the real code path," which serve as behavioral boundaries.
Capability inventory: The agent can execute shell commands (sh), file system operations via git, and network requests via gh and git fetch.
Sanitization: There is no automated sanitization of the code being reviewed; the skill relies on the agent's internal reasoning and explicit instructions to manually verify findings to prevent malicious code from influencing actions.

codex-review