The Agent Skills Directory

[SAFE]: The skill implements AES-256-GCM encryption for data at rest using standard libraries and secure parameters.\n
Evidence:\n
Analysis of scripts/agent.py and scripts/process.py shows proper use of PBKDF2-HMAC-SHA256 with 600,000 iterations for key derivation.\n
Cryptographically secure random nonces are generated via os.urandom() and are never reused for the same key.\n
The implementation includes integrity checks via the GCM authentication tag, ensuring that modified ciphertext will be detected during decryption.\n- [INDIRECT_PROMPT_INJECTION]: The skill possesses a potential attack surface for indirect prompt injection as it reads and processes external files.\n
Ingestion points: The encrypt_file and decrypt_file functions in scripts/agent.py and scripts/process.py read data from user-specified file paths.\n
Boundary markers: No explicit delimiters or instructions are used to distinguish between processed data and agent instructions.\n
Capability inventory: The skill is capable of writing files and creating directories on the local file system.\n
Sanitization: The implementation uses authenticated encryption (AES-GCM), which verifies the integrity of the encrypted data, preventing the processing of tampered ciphertext.

implementing-aes-encryption-for-data-at-rest