The Agent Skills Directory

[PROMPT_INJECTION]: No instructions were found that attempt to override the AI agent's safety guidelines or bypass its core operational constraints. The instructions are focused entirely on the stated task of code review.
[DATA_EXFILTRATION]: The skill does not perform any network operations or access sensitive system files. It actually includes security-positive instructions to identify and report hardcoded secrets or sensitive data exposure in the code being reviewed.
[REMOTE_CODE_EXECUTION]: There are no patterns involving the download or execution of remote scripts or packages.
[COMMAND_EXECUTION]: The skill does not invoke any shell commands or attempt privilege escalation. It functions as a set of logical instructions for the AI's analysis process.
[INDIRECT_PROMPT_INJECTION]: While the skill is designed to ingest and process untrusted data (user-provided code and diffs), it does not have access to exploitable capabilities like file system writes or network access. Furthermore, the skill's own logic includes checks for malicious patterns (like injections) in the data it processes, serving as a defensive layer.

code-review