The Agent Skills Directory

[PROMPT_INJECTION]: The skill's reference materials include several common injection strings, such as "Ignore previous instructions" and "reveal your system prompt". These are confirmed false positives; they appear within documentation for creating automated test suites (INJECTION_TESTS) to evaluate the robustness of LLM prompts against adversarial attacks.
[EXTERNAL_DOWNLOADS]: Reference files document the use of established third-party libraries and SDKs, including anthropic, openai, tiktoken, and pydantic. These tools are necessary for the implementation of the prompting and evaluation workflows described in the skill.
[COMMAND_EXECUTION]: Python examples provided in the skill demonstrate routine file system tasks, such as loading prompt templates and storing JSON evaluation reports. These operations are performed using standard libraries (pathlib, json) and are appropriate for the skill's intended development and testing functions.

prompt-engineer