memory-graph-builder

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The skill reads arbitrary lines from MEMORY.md and generates/writes a compressed digest including those memory texts verbatim (and logs summaries), so if secrets are stored in MEMORY.md the LLM would be required to output them without any redaction—creating an exfiltration risk.