The Agent Skills Directory

[COMMAND_EXECUTION]: The test suite scripts/test_draw_cards.py uses subprocess.run to call the local Python script for functional verification. This is handled safely by using the specific Python interpreter and direct file paths without invoking a shell, preventing command injection vulnerabilities.
[INDIRECT_PROMPT_INJECTION]: The skill introduces randomized content intended to influence agent behavior when prompts are vague or nonchalant. This design pattern is mitigated by explicit instructions in SKILL.md and the interpretation guides that require human-verifiable evidence and standard engineering best practices to take precedence over card readings in security and correctness contexts.
[SAFE]: The core logic in scripts/draw_cards.py correctly implements cryptographically secure randomness using the secrets module to generate its entropy budget, ensuring that the draws are unpredictable and tamper-resistant.

let-fate-decide