The Agent Skills Directory

[NO_CODE]: The skill consists entirely of natural language instructions and organizational guidelines. It does not contain any executable scripts, binary files, or automated commands.
[DATA_EXPOSURE]: The skill includes explicit security instructions to redact secrets and avoid inspecting sensitive files such as .env, .key, and .pem. It mandates that all experiments be confined to a specific scratch directory (.agents/scratch/).
[COMMAND_EXECUTION]: While the skill describes an execution protocol for running experiments, it does not provide any hardcoded shell commands or automated execution logic. It serves as a behavioral guide for the agent when tasked with validating solutions.

spike