The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes several command-line tools including git, gh, and npm. Specifically, it executes npm run preflight after checking out a remote Pull Request. If a Pull Request contains malicious modifications to the project's build or test scripts (e.g., in package.json), the agent will inadvertently execute arbitrary code provided by the external contributor.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) when reviewing untrusted external content. Ingestion points: The agent reads code changes via git diff and GitHub Pull Request descriptions/comments. Boundary markers: The skill lacks delimiters or explicit instructions for the agent to ignore embedded commands within the code being analyzed. Capability inventory: The skill has the ability to read the local filesystem and execute shell commands. Sanitization: There is no evidence of sanitization or filtering applied to the code or comments before they are processed by the LLM.
[REMOTE_CODE_EXECUTION]: By combining gh pr checkout (download) with npm run preflight (execution) on content from potentially unknown external authors, the skill creates a pathway for remote code execution if the Pull Request is malicious.

code-reviewer