The Agent Skills Directory

[SAFE]: The skill provides a structured methodology for Evaluation-Driven Development (EDD), focusing on defining success criteria and running regression tests.
[COMMAND_EXECUTION]: The framework utilizes standard shell commands such as grep, npm test, and npm run build for deterministic code-based grading. These operations are restricted to the local development environment and reflect standard engineering practices.
[DATA_EXFILTRATION]: There are no network-enabled commands or operations that attempt to access or exfiltrate sensitive data from the system.
[EXTERNAL_DOWNLOADS]: The skill does not perform any remote downloads or execute scripts from external sources.
[PROMPT_INJECTION]: The instructions are purely technical and functional; they do not contain any patterns designed to bypass AI safety guardrails or extract system prompts.

eval-harness