The Agent Skills Directory

[SAFE]: The skill is designed with security in mind, explicitly instructing the agent to use single-quoted heredocs (cat <<'ROBOREV_COMMENT') when passing review-derived data to shell commands. This practice prevents shell expansion and command injection from potentially malicious content found in code reviews.- [COMMAND_EXECUTION]: The skill performs standard development tasks, including running the roborev CLI tool and project test suites (e.g., go test). These actions are consistent with the skill's primary purpose of fixing code findings.- [PROMPT_INJECTION]: The instructions provide clear boundaries and safety guidelines, ensuring the agent remains focused on the code review task and uses secure methods for data processing.

roborev-refine