The Agent Skills Directory

[COMMAND_EXECUTION]: The run.sh script performs several command-line operations using git (such as rev-parse, merge-base, ls-files, diff, and log) to gather repository state and invokes the codex CLI tool to perform the code review.
[PRIVILEGE_ESCALATION]: The SKILL.md file advises users to use sudo when installing the @openai/codex package globally, which involves granting the installation process administrative system permissions.
[EXTERNAL_DOWNLOADS]: The skill requires the user to install the @openai/codex package from the official npm registry, which is a well-known service provided by a trusted organization.
[INDIRECT_PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it reads arbitrary file content and diffs from the local repository and includes them in a prompt for an external LLM.
Ingestion points: The run.sh script reads the contents of REVIEW.md, the output of git diff, and the full contents of all untracked files identified by git ls-files.
Boundary markers: There are no explicit delimiters or specific instructions in the constructed prompt to isolate the code being reviewed and prevent the LLM from following instructions embedded within that code.
Capability inventory: The script executes the codex CLI tool, which communicates with an external LLM service to process the review prompt.
Sanitization: The skill does not perform any sanitization or validation of the file contents or git diff data before appending it to the prompt.

local-review-codex