The Agent Skills Directory

[SAFE]: The skill follows a structured and transparent workflow for analyzing local features, incorporating a mandatory discovery phase that requires user input before analysis proceeds.
[COMMAND_EXECUTION]: The instructions describe using common filesystem search tools (grep and glob) to identify files relevant to the feature being analyzed, which is standard behavior for a development-focused agent skill.
[PROMPT_INJECTION]: While the skill reads codebase files that could potentially contain indirect prompt injections, the requirement for manual user discovery and the analytical context of the task reduce the risk of the agent autonomously executing instructions found within project data.

elite-feature