The Agent Skills Directory

[DATA_EXPOSURE]: The skill instructions direct the agent to perform an inventory of the live system, including reading the current repository state and configuration files located in the ~/.agents/skills directory. This is used to maintain context and ensure the agent does not duplicate existing work.
[PROMPT_INJECTION]: The protocol includes logic for parsing user input, specifically handling noise from voice-dictated transcriptions and prioritizing later corrections. These are instructional guidelines for intent recognition and do not represent a bypass of safety filters.
[COMMAND_EXECUTION]: The skill focuses on gathering information and thinking. It explicitly commands the agent to avoid implementing code or making file edits while this protocol is active, restricting its actions to read-only context gathering and verbal output.

think4ming