The Agent Skills Directory

[SAFE]: No malicious patterns or security vulnerabilities were detected in the provided documentation and instructions.- [PROMPT_INJECTION]: The instructions focus entirely on technical implementation details for deterministic execution. No bypass markers, DAN-style roleplay, or instructions to ignore safety protocols were found.- [CREDENTIALS_SAFE]: The documentation does not contain hardcoded API keys, secrets, or sensitive credentials. It follows best practices by treating seeds as explicit inputs rather than hardcoded implementation details.- [EXTERNAL_DOWNLOADS]: The skill references and pins specific versions of well-known, trusted libraries including NumPy, PyTorch, and TensorFlow. These references are documented for the purpose of ensuring environment reproducibility and do not involve untrusted remote code execution.- [OBFUSCATION]: Analysis of the markdown content and code snippets found no evidence of Base64 encoding, zero-width characters, homoglyph attacks, or hidden text patterns intended to conceal malicious intent.- [DATA_EXFILTRATION]: No unauthorized network operations or patterns for harvesting sensitive local files (such as SSH keys or environment variables) were detected.- [INDIRECT_PROMPT_INJECTION]: The skill addresses the ingestion of external data (snapshots and recorded inputs) by prescribing strict boundary controls, including canonical state encoding and cryptographic hash validation at defined compare-points to detect divergence.

using-determinism-and-replay