The Agent Skills Directory

[SAFE]: The skill consists of documentation and Python scripts designed for static analysis of prompts, RAG datasets, and agent configurations. All scripts (agent_orchestrator.py, prompt_optimizer.py, rag_evaluator.py) rely exclusively on the Python standard library (argparse, json, re, pathlib, dataclasses).
[COMMAND_EXECUTION]: While the skill documentation provides examples of CLI commands (e.g., python scripts/prompt_optimizer.py prompt.txt), the scripts themselves only perform local file reads and text processing. They do not invoke shell commands via os.system or subprocess.
[DATA_EXFILTRATION]: There are no network-capable libraries (like requests or urllib) or network operations detected in the source code. All analysis is performed locally on the machine where the scripts are run.
[PROMPT_INJECTION]: The skill documents defensive prompt engineering patterns, such as structured output and few-shot examples, to mitigate instruction bypass. It does not contain any malicious instructions or bypass attempts.
[CREDENTIALS_UNSAFE]: The provided code contains no hardcoded secrets or sensitive keys. References to API keys in documentation (e.g., in agent.yaml examples) are purely illustrative and used to explain configuration validation logic.

senior-prompt-engineer