The Agent Skills Directory

[SAFE]: The skill is primarily instructional and defines a specific communication style ('Caveman mode') for token reduction. It does not contain scripts or logic that interact with the system or network.
[SAFE]: No hardcoded credentials, sensitive file paths, or network exfiltration patterns were detected.
[SAFE]: The skill explicitly excludes security warnings and confirmation prompts from its compression rules, which is a safety best practice to ensure clarity during critical operations.
[SAFE]: Although the frontmatter includes powerful tools like Bash and Write in the allowed-tools list, this is a restrictive platform configuration that limits the agent's pre-existing capabilities rather than granting new ones, and no malicious command execution patterns are present in the skill body.

caveman