The Agent Skills Directory

[SAFE]: The skill acts as an orchestrator, dispatching other internal skills within the same plugin using the Skill() tool. This follows standard delegation practices.
[COMMAND_EXECUTION]: Executes local Python scripts (pipeline-summary.py, run-metrics.py, knowledge-binding.py, control-path.py) located within the plugin's root directory (${CLAUDE_PLUGIN_ROOT}/scripts/) for project tracking and reporting. These are internal vendor utilities used for metadata management.
[SAFE]: Implements a robust 'cost gate' mechanism using AskUserQuestion to ensure the user confirms potentially expensive operations (such as high-volume web research or data ingestion fan-out) before they execute.
[SAFE]: Data processing is limited to reading and writing local project metadata in JSON format and logging run outcomes to the wiki's activity log. No unauthorized data access or exfiltration patterns were detected.

knowledge-run