The Agent Skills Directory

[SAFE]: The skill instructs the agent to follow essential security practices when documenting code, such as using parameterized queries and avoiding hardcoded credentials. It also provides a detailed reference on common LLM documentation pitfalls, including security vulnerabilities.\n- [COMMAND_EXECUTION]: The workflow includes running the claude CLI to benchmark the model's baseline knowledge. This is a legitimate diagnostic step intended to reduce documentation redundancy.\n- [COMMAND_EXECUTION]: The agent is directed to read local project files and use a Task tool to invoke specialized sub-agents (researcher and docs-reviewer) for technical accuracy checks and content optimization. These actions are performed within the scope of documentation development.

writing-documentation