The Agent Skills Directory

[SAFE]: The skill manages persistent memory through a specific MCP server provided by the vendor. It follows standard practices for data persistence and does not access sensitive system resources, credentials, or unauthorized external domains.
[SAFE]: Analysis of indirect prompt injection surfaces confirms that the skill is designed to process untrusted data as its primary function, but it lacks exploitable capabilities that would elevate the risk.
Ingestion points: User input enters the system via memory_add and memory_update tools, and is subsequently retrieved through memory_search as described in SKILL.md and references/memory-guide.md.
Boundary markers: The instructions do not specify the use of clear delimiters or instructions to ignore embedded content when the agent processes retrieved memories.
Capability inventory: The agent's tools are strictly limited to the mcp_datadata_memory_* suite for CRUD operations on memories. There is no access to shell execution, arbitrary file system operations, or network exfiltration tools.
Sanitization: No explicit sanitization or input validation logic is defined in the provided skill instructions or reference guides.

datadata-memory