The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local system commands including jq, rm, and mv via a Bash script to manage JSON configuration files. These operations are confined to the .claude subdirectory of the project and are used for legitimate configuration management purposes.\n- [DATA_EXFILTRATION]: No network activity or external data transmission mechanisms were detected. Data remains within the local environment and is only displayed to the agent.\n- [EXTERNAL_DOWNLOADS]: The skill does not attempt to download external packages or remote scripts. It utilizes common system utilities like jq and python3 which are expected to be pre-installed.\n- [PROMPT_INJECTION]: There are no instructions or patterns in the skill metadata or script body that attempt to bypass safety guidelines, override system instructions, or extract sensitive prompt data.

config