The Agent Skills Directory

[SAFE]: No security threats detected. The skill focus is educational, providing instructions on hardening LLM prompts against adversarial inputs and enforcing structured outputs.
[EXTERNAL_DOWNLOADS]: The skill mentions and provides integration examples for standard libraries including openai, anthropic, pydantic, jsonschema, and dspy. These are well-known, industry-standard packages for AI development and data validation.
[COMMAND_EXECUTION]: The scripts/verify.sh script is a utility for linting prompt files. It is designed to be read-only and does not perform network operations or write to the filesystem, adhering to safety best practices.
[PROMPT_INJECTION]: The skill specifically instructs users on how to mitigate prompt injection by using delimiters for untrusted input and phrasing constraints positively. It does not contain any malicious injection attempts against the agent.

prompt-engineering