The Agent Skills Directory

[Data Exposure] (LOW): The skill accesses ~/.neuralmemory/config.toml as part of its context. While specific to the application, local configuration files can contain sensitive metadata or environment tokens.- [Indirect Prompt Injection] (LOW): The skill processes untrusted external data in the form of system memories, creating a potential surface for indirect injection.
Ingestion points: Memory content is ingested via nmem_recall and nmem_context.
Boundary markers: Absent; the skill does not use delimiters to isolate memory content from its own instructions.
Capability inventory: The skill uses diagnostic tools only (nmem_stats, nmem_health, nmem_conflicts). No file-write, network exfiltration, or shell execution tools are available to the agent.
Sanitization: None; memory content is analyzed directly to produce the audit report.

memory-audit