The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides a utility script, scripts/validate-agent.sh, which uses standard shell commands (grep, sed, awk, head, tail) to parse and validate the syntax of agent markdown files. This is a functional utility for local development and does not perform network operations or access sensitive system paths.
[SAFE]: The documentation consistently recommends secure patterns such as the principle of least privilege when defining agent tool access. The example agents (code reviewer, test generator, docs generator, and security analyzer) are structured with clear responsibilities, quality standards, and restricted toolsets.
[PROMPT_INJECTION]: There are no detected attempts to bypass agent safety guidelines. The instructional templates provided in the references are meta-prompts designed to help users generate well-formatted JSON and markdown for their own sub-agents, adhering to project-specific standards.

agent-development