deep-agents-memory

Pass

Audited by Gen Agent Trust Hub on Oct 28, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • Filesystem Management Tools: The skill documentation includes tools for file system interaction, such as read_file, write_file, and edit_file. These capabilities allow agents to manage workspace data and are presented alongside security recommendations like using virtual_mode to enforce root directory restrictions and prevent directory traversal.
  • Indirect Prompt Injection Surface: This skill enables agents to read data from external files, which creates a potential surface where untrusted content could influence agent behavior. The skill documentation explicitly mitigates this by teaching developers to use restricted execution environments and implementation of interrupt_on triggers for human oversight of file changes.
  • Persistence Configuration Patterns: The skill describes patterns for long-term memory using databases like PostgreSQL. These implementation guides use standard connection string templates for configuration and are consistent with established architectural practices for persistent agent state.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 28, 2026, 05:18 AM