The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is configured to use standard development tools through Bash(git:*), Bash(yarn:*), and Bash(npm:*). These are used for their intended purpose of checking, linting, and building code during a review process.
[INDIRECT_PROMPT_INJECTION]: As a code review tool, the skill processes untrusted external data in the form of code diffs and local files.
Ingestion points: Processes code diffs and local project files via the Read, Grep, and Glob tools.
Boundary markers: Not explicitly defined in this entry-point skill file.
Capability inventory: The skill can execute shell commands (bash), manage packages (npm, yarn), and interact with version control (git).
Sanitization: Standard linting and building processes provide a layer of validation, though the skill relies on the underlying agent's safety guardrails when processing file content.

codex-review