The Agent Skills Directory

[PROMPT_INJECTION]: The skill documents an attack surface for indirect prompt injection through tools that process external or historical content.
Ingestion points: memory_signal_read (reads inter-agent messages) and memory_recall/memory_smart_search (retrieve historical observations) are primary data ingestion vectors.
Boundary markers: The reference material lacks specific instructions or delimiters to isolate ingested data from agent instructions.
Capability inventory: The toolset includes file system modification (memory_compress_file, memory_obsidian_export), state synchronization (memory_mesh_sync), and data export capabilities (memory_export).
Sanitization: No mention of content sanitization or validation is present in the tool registry documentation.
[COMMAND_EXECUTION]: The memory_compress_file tool performs local file operations by creating backups and rewriting markdown files to optimize token usage.
[DATA_EXFILTRATION]: The skill provides instructions for tools that move data across boundaries, including memory_export for dumping state to JSON and memory_mesh_sync for sharing memory with peer agent instances.

agentmemory-mcp-tools