The Agent Skills Directory

[SAFE]: The skill implements a robust 'L4 Core Layer Protection Principle' which explicitly forbids the AI from writing to files containing core values, beliefs, or identity definitions. This prevents automated tampering with the agent's fundamental persona or the user's defined ethics.
[SAFE]: File system operations are restricted to specific local paths (AI_MEMORY/) and use standard markdown formatting. There are no network requests, remote script executions, or credential exposures.
[SAFE]: The provided Python script (scripts/update_memory.py) is used for local formatting and path calculation; it does not perform any dangerous system calls or external data fetching.
[COMMAND_EXECUTION]: The skill uses grep for pattern detection across memory files. While this involves shell execution, it is limited to searching the local memory directory for statistics and does not incorporate unvalidated user input into the command structure in a way that allows shell injection.

mem-record