The Agent Skills Directory

[SAFE]: The skill utilizes the Gemini CLI, which is an official tool provided by a well-known service provider. Downloads and interactions with this tool are considered standard professional practice.
[SAFE]: The instructions explicitly mandate the use of the --approval-mode plan flag during execution. This provides a critical security boundary by ensuring the external CLI agent cannot perform unauthorized file modifications or write operations to the local environment.
[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by ingesting untrusted external data (project diffs and logs) and interpolating it into a prompt template for an external LLM. An attacker could embed instructions within a code change to influence the resulting review, though the impact is limited by the read-only execution flags and the final requirement for human review.

gemini-review