The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill follows recommended security practices by using environment variables (e.g., os.environ['POSTGRES_URL'], os.environ['PINECONE_API_KEY']) to manage sensitive connection strings and API keys rather than hardcoding them in scripts.
[EXTERNAL_DOWNLOADS]: The skill references established and reputable technology services including Pinecone, Qdrant, ChromaDB, and LangChain. No downloads from untrusted or unknown third-party sources were detected.
[SAFE]: The documentation explicitly identifies security risks inherent in memory systems, such as data leakage between users, and provides actionable advice for developers to implement metadata filtering and validation checks.
[SAFE]: No obfuscation, persistence mechanisms, or unauthorized privilege escalation commands were found within the skill instructions or code snippets.
[PROMPT_INJECTION]: The skill contains an inherent attack surface for indirect prompt injection as it processes user conversation data to extract insights for memory storage. However, the skill provides architectural guidance on using LLMs for consolidation and conflict resolution, which is standard for this use case, and it does not contain any malicious injection patterns.

agent-memory-systems-v2