The Agent Skills Directory

[COMMAND_EXECUTION]: The skill accepts a user-provided 'fixed point' (such as a branch name, tag, or commit SHA) and incorporates it directly into shell commands including git diff, git log, and git rev-parse. This pattern presents a potential command injection risk if the user input contains shell metacharacters and the execution environment does not properly sanitize these arguments.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection. It reads untrusted content from the repository (code diffs) and external sources (PRDs, specs, and issue tracker entries) and includes this data in the instructions sent to sub-agents. The sub-agent prompts lack explicit boundary markers or 'ignore' instructions to prevent the sub-agents from being influenced by malicious instructions embedded within the code or documentation being reviewed.

code-review