AgentDB Memory Patterns

[Skill Scanner] Backtick command substitution detected This skill documentation and examples describe a local persistent memory subsystem for agents (AgentDB + ReasoningBank) and are internally consistent with their stated purpose. There is no direct evidence in the provided text of obfuscated or malicious code, hardcoded secrets, or explicit exfiltration. The main security concerns are operational: (1) npx package installs and MCP/Claude integration introduce external trust dependencies that must be validated, and (2) migration commands that read legacy DB paths could unintentionally copy sensitive data. Recommend reviewing the actual agentdb/agentic-flow package code and the MCP integration implementation before use, and verifying the trustworthiness of external domains (agentdb.ruv.io, GitHub repo) and the npm package publisher. LLM verification: BENIGN: The Skill.md fragment coherently describes memory patterns, persistence, and ReasoningBank integration. No malicious behavior detected within the provided content. Treat performance claims as marketing language pending independent benchmarks. Minor documentation-time anomalies (template literals) are non-executable in this context.