The Agent Skills Directory

[NO_CODE]: The skill consists entirely of documentation and architectural patterns. No executable scripts, command-line operations, or network calls are present.
[INDIRECT_PROMPT_INJECTION]: The skill discusses ingestion of external data for memory (RAG, Vector DBs). However, it does not include any exploitable scripts and provides theoretical guidance on filtering and ranking to prevent 'Context Poisoning', which is a security best practice for managing memory layers.
[SAFE]: References to external frameworks like Mem0, Cognee, and Letta are standard in the industry for agent memory management and are mentioned as integration possibilities, not as immediate unverified downloads.

memory-systems